Paper discovery
植物形質を測っただけの研究ではなく、フェノタイピング手法の開発・検証・実質的利用・ベンチマーク・方法レビューとの関連性が見つかった論文を中心に表示します。
Abstract has not been obtained from indexed metadata or an accessible article page.
Agriculture plays a pivotal role in ensuring global food security, economic stability, and sustainable development. Plant diseases significantly reduce agricultural productivity, resulting in substantial economic losses and threatening food supply worldwide. Early and accurate identification of plant leaf diseases enables timely intervention, minimizes crop damage, and enhances agricultural yield. Traditional disease diagnosis relies heavily on visual inspection by agricultural experts, making the process labor-intensive, subjective, and unsuitable for large-scale deployment. Recent advances in artificial intelligence, particularly deep learning, have transformed plant disease diagnosis by enabling automatic feature extraction and highly accurate image-based classification. This review presents a comprehensive analysis of recent developments in deep learning techniques for plant leaf disease identification. Various convolutional neural network (CNN) architectures, including AlexNet, VGGNet, ResNet, DenseNet, EfficientNet, MobileNet, Inception, and Xception, are critically reviewed along with modern transformer-based models such as Vision Transformer (ViT), Swin Transformer, and hybrid CNN–Transformer frameworks. The paper also examines transfer learning strategies, object detection methods including YOLO and Faster R-CNN, and semantic segmentation approaches such as U-Net and DeepLabV3+. Publicly available benchmark datasets, including PlantVillage, PlantDoc, AI Challenger, Cassava Leaf Disease, and Rice Leaf Disease datasets, are discussed in terms of dataset diversity, annotation quality, and practical applicability. Furthermore, image preprocessing techniques, data augmentation methods, evaluation metrics, and deployment considerations for mobile and edge devices are comprehensively reviewed. The paper identifies current research challenges, including dataset imbalance, environmental variability, model interpretability, computational complexity, and limited real-world generalization. Finally, emerging research directions such as explainable artificial intelligence, federated learning, multimodal learning, self-supervised learning, lightweight architectures, and edge AI are discussed to provide future research opportunities. This review serves as a valuable resource for researchers, practitioners, and agricultural technologists interested in developing robust, scalable, and intelligent plant disease identification systems.
Cassava brown streak disease (CBSD) threatens food security for millions in East Africa, yet its control remains limited by the absence of field-deployable molecular diagnostics. Here, we introduce ELLA (Electrochemical Lateral flow assay with Linked Analytics), a battery-free, smartphone-powered electrochemical lateral flow assay that delivers enzyme-linked immunosorbent assay (ELISA)-grade protein detection directly in the field. ELLA integrates near-field communication, a single-chip potentiostat, metal-pin electrodes, and ferrocene-labeled nanoparticles into a fully disposable cassette, enabling quantitative immunoassays without optical instrumentation or centralized laboratory infrastructure. Validated across laboratory studies and extensive field trials in Tanzania, ELLA achieved 95% agreement with ELISA and 89% agreement with RT-qPCR, outperforming ELISA’s limit of detection while maintaining a material cost below US$1. By coupling molecular test results with cloud-linked analytics, we further trained DeepELLA, a smartphone-based image classification model that enables scalable surveillance from field-acquired leaf images. Together, these advances unify electrochemical sensing, digital connectivity, and AI-assisted interpretation, enabling portable, ELISA-level diagnostics for plant, environmental, and health monitoring in resource-limited regions.
The acquisition of labelled data for new or emerging plant diseases is difficult due to the high cost and logistical complexity of ground-truth collection, challenges that are further compounded by the limited infrastructure available to smallholder farmers in sub-Saharan Africa. This study presents FewShotCropNet, a few-shot learning model based on Spectral-Temporal Attention Mechanisms and Prototypical Networks that utilises multispectral time-series data from Sentinel-2 to classify crop diseases. Two principal innovations are incorporated in the proposed model: (1) a spectral attention mechanism based on the Squeeze-and-Excitation approach to learn disease-relevant spectral band weights; and (2) a temporal attention pooling mechanism to identify the most discriminative growth stages for disease classification. The model employs a two-phase training strategy comprising supervised pre-training followed by episodic meta-learning, enabling the generation of optimal feature representations under extreme label scarcity. Crop disease detection experiments were conducted in Edo State, Nigeria on cassava and maize using monthly Sentinel-2 composites from 2024 (10 spectral bands and five vegetation indices across twelve temporal steps). Under a 4-way 5-shot classification scenario with 100 GPS-validated labelled samples (25 per-class), FewShotCropNet achieved a mean accuracy of 98.15% with a 95% confidence interval of ±0.58%. An equitable comparison was enabled by introducing a Pre-trained Simple Prototypical Network, a variant sharing the same two-phase training strategy as FewShotCropNet but without the attention modules—which achieved 97.75% (±0.63%). FewShotCropNet exceeded the Pre-trained Simple ProtoNet by +0.40 percentage points (t = 1.52, p = 0.13), with the attention module contribution verified as positive though not statistically significant on the current dataset. Statistically significant improvements over models trained without pre-training were observed: FewShotCropNet outperformed the Relation Network (94.40%), Matching Network (94.40%), and the Optimised Baseline Convolutional neural networks (CNN) (86.65%, pre-trained backbone with 5-shot linear probe), with p
Tapioca is a vital root crop whose yield is influenced by multiple environmental, soil, and physiological factors.Traditional yield estimation methods depend upon statistical models and manual sampling, which often lack real-time accuracy and are time-consuming and labour-intensive.The combination of Internet of Things (IoT) sensor networks and Unmanned Aerial Vehicle (UAV)/satellite imagery, with deep learning (DL) models, gives a promising solution for accurate real-time yield prediction.In this study, sensor data (soil moisture, weather parameters, pH, NPK levels, temperature, and leaf chlorophyll content) are collected using devices like Davis Vantage Pro2, Decagon 5TE and SPAD-502, while image data are captured using DJI Phantom 4 Multispectral UAVs and Sentinel-2 satellite imagery.Pre-processing of sensor data involves missing value imputation, normalization, and feature selection using the Hybrid Frilled Lizard Osprey (HFLO) algorithm, while image data undergo resizing, augmentation, and filtering approach.Image based features are extracted by a position-attention DenseNet-201 model.Also, the features from image data and sensor data are fused using a concatenation mechanism.Finally, fully connected layers and improved support vector regression (ISVR) are used to identify the tapioca yield prediction.The integrated DL methods give higher accuracy than individual modalities for both modalities.The image-based model captures spatial and phenotypic variations, while the sensor-based model captures fine-grained environmental effects.The proposed approach obtained the MAE value of 0.0566, RMSE value of 0.654, and R 2 value of 99.5, demonstrating the potential of multimodal real-time data integration for precision agriculture in tapioca fields.
Abstract Maize and cassava are staple crops in Nigeria, but their productivity is limited by viral and fungal diseases. This study created a mobile system based on lightweight CNN for smartphone portable real-time detection of cassava and maize diseases. In the 2025 planting season, 10,800 leaf images were gathered from a 10-acre experimental farm in Araromi Area, Bakatari Farm, Ido Local Government, Ibadan that cut across the seven classes of healthy cassava (1,620, 15%), cassava mosaic disease (1,540, 14.3%), cassava brown streak disease (1,410, 13.1%), healthy maize (1,880, 17.4%), maize leaf blight (1,540, 14.3%), maize rust (1,360, 12.6%) and maize streak virus (1,450, 13.4%). In order to make the dataset more diverse, data augmentation was done in the form of rotation by +/-30°, flipping, brightness by +/-20% and random cropping in the range of 80-100%. Lightweight CNN architectures MobileNetV2, EfficientNet-Lite, ShuffleNet, and custom CNN were trained in an 80:20 ratio for train and test. Out of 11 models tested, EfficientNet-Lite model forecast the highest where it achieved an accuracy of 94.6%, precision of 0.95, recall of 0.94, F1-score of 0.94, and an ROC-AUC of 0.97. As for MobileNetV2, it achieved an accuracy of 93.8% while ShuffleNet was estimated to achieve the fastest mobile inference at 65 ms. As for the class-wise analysis, it can be seen that the maize leaf blight (95.2%) and cassava mosaic disease (94.1%) had the most accurate predictions. The offline prediction from mobile deployment showed that EfficientNet-Lite occupied 92 ms and 135 MB. The findings show that low-cost and practical smartphone-based disease diagnosis requires the use of lightweight CNN models that can deliver accuracy and alertness that can allow farmers to manage the crop and food security on their own. Smallholder farmers in a resource-poor rural environment will be able to benefit from these models.
Cassava is a major staple crop in tropical regions, particularly in Sub-Saharan Africa, yet its productivity remains constrained by genetic and agronomic limitations. A major bottleneck in cassava breeding is the difficulty of accurately phenotyping agronomic traits under field conditions using conventional, labor-intensive methods. Here, we evaluated the potential of uncrewed aerial vehicle (UAV)-based phenotyping to quantify canopy growth traits and assess their genetic relevance under realistic field conditions. For this, multi-temporal UAV imagery was collected over two growing seasons (2018-2019 and 2019-2020) in a panel of 46 cassava genotypes planted in fields of the International Institute for Tropical Agriculture (IITA), Nigeria. Canopy height, canopy volume, and their relative growth rates (RGRh and RGRv) were extracted at the plot-level, and their seasonal dynamics and canopy-yield relationships were further assessed across developmental stages and environmental conditions. Repeatability (R) and broad-sense heritability (H2) were estimated using a linear mixed model (LMM) that partitioned genetic, genotype-by-year, and residual variance components, enabling the evaluation of both measurement reliability and genetic signal. Overall, UAV-derived growth dynamics were found to exhibit comparable patterns across genotypes, reflecting shared seasonal growth trajectories, while canopy-yield relationships varied with developmental stage and environmental conditions. In terms of genetic metrics, R was high for all UAV-derived traits (R = 0.68-0.69), indicating reliable genotype-level assessment across replicates and seasons. In contrast, H2 differed substantially among traits. Canopy volume (H2 = 0.64) and canopy height (H2 = 0.58) exhibited moderate-to-high heritability, reflecting strong genotype effects and comparatively moderate genotype-by-year interactions. However, their relative growth rates showed near-zero H2 values, driven primarily by genotype-by-year interaction, indicating a dominant environmental influence. These results demonstrate that UAV-derived canopy height and volume provide a consistent basis for genetic differentiation of cassava genotypes across environments, supporting their use in selection, whereas growth-rate traits are better suited for characterizing growth plasticity and genotype-by-environment interactions.
Abstract has not been obtained from indexed metadata or an accessible article page.
Abstract In sustainable agriculture, detecting pests and diseases early is critical. Recent technological advances in deep learning (DL) and multimodal imaging like multispectral and thermal data crop health monitoring is promising. Despite the progress, obtaining high accuracy across various crops with real-time performance is still a challenge. The hybrid convolutional neural network (CNN)-attention model integrating multispectral and thermal data for pest and disease detection has been introduced. A total of 1760 samples were collected from six crops (maize, rice, wheat, tomato and cassava), across different growth stages, labelled fungal, bacterial, viral and pest infections. The data was divided into 70% training, 15% validation, and 15% test sets. 3,500 samples were used for training. 750 samples were used for validation and test set. The hybrid CNN-attention model was contrasted with certain baseline models (SVM, Random Forest, CNN-RGB, CNN-Multispectral) and certain fusion methods (early, late, and hybrid fusion) based on accuracy, precision, recall, F1-score, and early detection sensitivity. The highest accuracy of 91.0% for rice at the vegetative stage was achieved by the hybrid model. It beats baseline and fusion models. The F1-score of the classification was reasonably high. Rice's sensitivity is 88.1%, and maize is 87.3%. The model fared well for all classes, getting 92.0 % for the healthy plant and 88.2 % for pest infestation. Future work can enhance the dataset with more crops and diseases and environmental factors and optimize detection time and early sensitivity for real-time deployment in agricultural decision support systems.
Deep learning has improved automated plant disease detection by increasing recognition accuracy and robustness compared with traditional vision-based methods. Self-supervised learning (SSL) further reduces dependence on manual labels, but its transferability across heterogeneous agricultural datasets remains insufficiently characterized. Here, we evaluate a contrastive SSL pretraining and fine-tuning pipeline, termed PlantCLR, for plant disease classification under cross-dataset transfer with target-domain fine-tuning. PlantCLR combines SimCLR-style contrastive pretraining with a lightweight convolutional classifier to balance representation quality and deployment efficiency. Experiments on PlantVillage and Cassava Leaf Disease show strong performance, achieving 99.10% accuracy and 99.04% F1-score on PlantVillage, and 96.83% accuracy and 96.70% F1-score on Cassava. Feature embedding visualization using t-SNE and explanation maps using Grad-CAM indicate improved class separability and attention to disease-relevant regions. These results suggest that contrastive SSL can improve representation transfer while maintaining computational efficiency, supporting scalable plant disease diagnostics in practical agricultural settings. Code is available at GitHub .
Abstract Cassava diseases such as Cassava Mosaic Disease (CMD), Cassava Brown Streak Disease (CBSD), and Cassava Bacterial Blight (CBB) pose serious threats to global food security, particularly in resource-limited regions where expert diagnosis is scarce. Although large vision–language models enable automated plant disease recognition, existing fine-tuning approaches struggle under extreme data scarcity. This paper proposes Progressive Layer Activation CLIP (PLA-CLIP), a curriculum-inspired fine-tuning framework for efficient few-shot classification of cassava diseases. PLA-CLIP progressively unfreezes transformer layers during training, stabilizing the optimization process while preserving pretrained vision–language alignment. Using only 43 images per class, PLA-CLIP achieves 78.25% accuracy and a 78.00% F1-weighted score on CD1, outperforming zero-shot CLIP by +15.98% and standard fine-tuning by +3.94%. Cross-dataset evaluations on CD2 and CD3 demonstrate robust generalization across varying conditions. Attention map visualizations confirm that the model focuses on disease-relevant regions, supporting interpretability. With a 2.65 ms inference time and moderate model size, PLA-CLIP offers an effective balance between efficiency and performance for practical plant health monitoring. The implementation and experimental code are publicly available at https://github.com/ mshafay5/PLA-CLIP.
Deep learning provides new methods for crop pest and disease identification and control, offering unique advantages in terms of recognition accuracy and efficiency. However, deep learning models generally lack interpretability, and their internal decision-making processes are difficult to understand. This, to some extent, undermines users' trust in the model's predictions and hinders its large-scale application in agricultural production. Therefore, improving model transparency and interpretability has become an important research direction. To address this issue, this study proposes a novel interpretable crop pest and disease identification model, the Contrastive Prototype Tree (CPTR). The model is designed around the core structure of "concept prototypes and decision tree," which builds clear prototype matching paths for each recognition result. This enables the model to not only have strong classification capability but also provide intuitive explanations. Additionally, the study introduces the SimCLR contrastive learning framework to enhance the model's ability to express deep image features. SimCLR guides the model to learn more discriminative visual features by maximizing the similarity between positive sample pairs and minimizing the similarity between negative sample pairs, thereby improving overall recognition performance. This study evaluated the model on three datasets: AppleLeaf9, Cassava, and Cashew. The experimental results show that CPTR achieves accuracies of 83.74%, 94.80%, and 96.01% on the three datasets, representing improvements of 4.12%, 0.34%, and 0.51% compared to Prototype Tree, respectively. These results indicate that the proposed model achieves the highest accuracy across different datasets, demonstrating its effectiveness.
Monitoring the growth condition of cassava plants is a crucial aspect of agriculture to ensure optimal yields. The manual identification process is often subjective, time-consuming, and requires special expertise. This research aims to develop an automatic classification model using the Convolutional Neural Network (CNN) method to identify the growth conditions of cassava plants based on leaf imagery. A dataset of cassava leaf images with three categories Healthy, Stressed, and Diseased was used to train and test the CNN architecture. The research methodology includes data collection, image augmentation to enrich data variance, model architecture design, training, and evaluation. The results indicate that the developed CNN model achieves good accuracy in distinguishing among the three leaf condition classes. This system has the potential to assist farmers in detecting plant health issues early and accurately, thereby supporting more effective decision-making in agricultural management.
This study explores the use of hyperspectral imaging (HSI) combined with machine learning to detect physiological alterations in cassava leaves caused by Xanthomonas phaseoli pv. manihotis (Xpm), a bacterial plant disease that causes significant yield losses worldwide. Therefore, the use of hyperspectral images associated with machine learning can provide information rapidly and accurately, aiming to support decision-making. HSI captures spectral data that reflects biochemical changes in infected plant tissues. An image set of cassava healthy and symptomatic leaves (402 and 450, respectively) were imaged using a hyperspectral camera across wavelengths from 400 to 1000 nm, with image calibration and spectral normalization to improve data quality. Spectral parameters, such as mean reflectance and spectral differences (healthy vs. infected), were analyzed. Six machine learning models were tested for classification: Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Extreme Gradient Boosting (XGBoost), and Multi-Layer Perceptron (MLP). SVM performed best, achieving the highest accuracy (91.41%), followed by MLP (87.89%), XGBoost (79.69%), and RF (77.34%). DT and KNN had the lowest accuracy (71.88% and 70.31%, respectively). The results suggest that HSI, particularly when combined with SVM, offers a rapid and accurate method for diagnosing cassava bacterial blight, with potential for large-scale field applications.
Cassava remains a critical food-security crop across Africa and Southeast Asia but is highly vulnerable to diseases such as cassava mosaic disease (CMD) and cassava brown streak disease (CBSD). Traditional diagnostic approaches are slow, labor-intensive, and inconsistent under field conditions. This review synthesizes current advances in combining unmanned aerial vehicles (UAVs) with deep learning (DL) to enable scalable, data-driven cassava disease detection. It examines UAV platforms, sensor technologies, flight protocols, image preprocessing pipelines, DL architectures, and existing datasets, and it evaluates how these components interact within UAV–DL disease-monitoring frameworks. The review also compares model performance across convolutional neural network-based and Transformer-based architectures, highlighting metrics such as accuracy, recall, F1-score, inference speed, and deployment feasibility. Persistent challenges—such as limited UAV-acquired datasets, annotation inconsistencies, geographic model bias, and inadequate real-time deployment—are identified and discussed. Finally, the paper proposes a structured research agenda including lightweight edge-deployable models, UAV-ready benchmarking protocols, and multimodal data fusion. This review provides a consolidated reference for researchers and practitioners seeking to develop practical and scalable cassava-disease detection systems.
Cassava leaf disease detection plays a critical role in safeguarding crop yield and supporting sustainable farming practices in regions where cassava is a primary food source. The task remains challenging due to visually similar disease symptoms, variability in lesion shapes and colors, inconsistent lighting conditions in field images, and overlapping infections. Existing deep learning and hybrid vision models, although effective in controlled environments, often suffer from high computational demands and limited capability to jointly capture fine lesion textures and long-range spatial relationships. To address these limitations, this research introduces Lite-RGA-GTNet, a lightweight residual graph-augmented graph-transformer network with spectral-directional preprocessing and progressive token pruning. The approach integrates RGB data with directional gradient and vegetation index maps, employs residual graph reasoning before attention layers, and fuses local-global features through hierarchical graph-transformer modules to produce compact yet context-rich representations. Experimental evaluation on a benchmark cassava leaf image dataset, consisting of five classes including healthy and diseased samples, demonstrates that Lite-RGA-GTNet achieves 96.84% accuracy, 96.25% precision, 96.72% recall, and 96.48% F1-score, surpassing existing models such as CassNet and LeafXFormer by up to 2.65% in accuracy, while maintaining an average inference time of 14 ms-indicating its suitability for real-time agricultural deployment.
Introduction Cassava is one of the most widely cultivated crops worldwide, renowned for its rich natural ingredients and numerous nutritional benefits. However, the complex interdependencies among its features often pose challenges in image restoration and segmentation, particularly when identifying disease regions. In previous work, this manifested as higher false positives and misidentification of non-relevant areas, leading to a decline in precision and accuracy. Methods To address these issues, this study proposed an efficient artificial intelligence-powered image analysis system that leverages optimal feature selection with a HyperCapsInception-ResNet-V2-CNN model to enhance disease detection accuracy. Initially, the dataset was collected from the Kaggle repository, its name was Cassava Leaf Disease Classification, and it comprised 21,367 different images. Our approach began by normalizing cassava plant disease data using adaptive Gaussian Otsu thresholding. Histogram color evaluation and iterative clustering fragmentation were then applied to better isolate disease variations and improve precision. Subsequently, Cascaded Canny Edge Segmentation (CCES) was used to effectively segment the disease region. The disease variation properties were further evaluated using the Optimal Spider Swarm Intelligence Technique (OSSIT) to reduce irrelevant feature dimensions. For classification, the HyperCapsInception-ResNet-V2-CNN model was employed to categorize cassava diseases, including cassava bacterial blight (CBB), cassava mosaic disease (CMD), cassava green mite (CGM) disease, and cassava brown streak disease (CBSD), along with regular and abnormal leaf states. Results The proposed method's simulation results achieved 98.15% accuracy, a 97.22% F1-score, and 96.02% precision, outperforming other traditional methods such as EfficientNetB3, AlexNet, Faster-RCNN, and InceptionV3. Discussion Both optimized feature selection with OSSIT and hybrid HyperCapsInception-ResNet-V2-CNN architecture significantly enhanced the detection reluctance and the classification of the data. These findings indicate that the proposed system is effective in the automated detection of cassava disease and has a high potential of being practical in agricultural practices especially in precision farming and early detection of diseases.
Cassava (Manihot esculenta Crantz) is a staple food and a key industrial crop across tropical regions, but traditional phenotyping for critical quality traits like dry matter content (DMC) and starch content (StC) is a laborious and low-throughput process. This study investigates the efficacy of a handheld near-infrared spectrometer device (NIRS) for the non-destructive, rapid prediction of these traits. The research methodology involved collecting spectral data from 2,236 cassava clones from 19 field trials in Brazil, using two sample types: fresh roots and mashed roots. Six spectral pre-processing methods and three machine learning algorithms-Partial Least Squares (PLS), Support Vector Machines (SVM), and Extreme Gradient Boosting (XGB)-were evaluated to optimize predictive models. Model performance was assessed using the coefficient of determination in calibration ([Formula: see text]), the root mean squared error of calibration ([Formula: see text]), and the Kappa index to quantify the consistency of clone selection. Results show that mashed samples consistently yielded superior predictive performance across all models. Specific preprocessing methods, such as Savitzky-Golay filtering combined with Standard Normal Variate (SG + SNV) and first-derivative transformations, significantly enhanced model accuracy. Among the algorithms, PLS demonstrated the best overall performance, with high predictive accuracy ([Formula: see text] >0.96) and low prediction errors ([Formula: see text]<1.3 for DMCo), especially with mashed samples. High Kappa index values, consistently approaching 1.0, confirmed a good alignment between NIRS-based selection and traditional phenotypic methods. This study validates a portable spectrometer as a reliable and efficient tool for high-throughput phenotyping in cassava breeding programs. The findings confirm that portable NIRS devices, when used with optimal sample preparation (mashed roots) and robust modeling (PLS), can effectively yield good predictions for plant selection. This approach can significantly accelerate breeding cycles by enabling rapid, early-stage selection decisions, thereby overcoming a major bottleneck and contributing to a more efficient and sustainable genetic improvement of cassava.
Abstract Cassava represents a crucial staple crop in sub-Saharan Africa, whose productivity is severely threatened by viral and bacterial diseases. While deep learning approaches show promise for automated disease detection, they often operate under unrealistic assumptions and suffer from dataset limitations including class imbalance and label noise. This paper presents a novel hierarchical deep learning framework that addresses these challenges through a three-stage pipeline: crop verification, health assessment, and disease classification. Our approach achieves computational efficiency through conditional execution while maintaining diagnostic accuracy. On the Cassava Leaf Disease Classification dataset comprising 21,397 images, our framework achieves 91.8\% overall accuracy with statistically significant improvements over flat classification baselines (mean difference: 1.7\%, bootstrap 95\% CI: 1.2-2.2\%, p
Abstract Cassava ( Manihot esculenta Crantz) is an important food security crop in sub-Saharan Africa and other tropical regions, but its genetic improvement is hindered by long breeding cycles and labour-intensive phenotyping procedures. This study aimed to develop a rapid phenotyping protocol and assess its predictive capacity for yield and plant architecture traits in cassava using Normalized Difference Vegetation Index (NDVI) data obtained with affordable handheld sensor (Trimble GreenSeeker). A diverse panel of 453 cassava accessions was evaluated across two contrasting agroecological zones in Nigeria; Mokwa (Southern Guinea Savannah) and Onne (Humid Forest) during the 2021/2022 planting season. NDVI data collected at 3, 6, and 9 months after planting (MAP) were integrated with ground truth phenotypic measurements of 26 agronomic traits.Genetic parameters including broad-sense heritability and genotype-by-environment interactions were estimated. Results showed moderate to high heritability for important traits such as fresh root yield (FYLD), dry matter content (DM), and harvest index (HI). NDVI data, especially at 6 months after planting, demonstrated strong predictive power (R² up to 0.9) for yield components, with prediction accuracy varying across locations. Significant negative correlations between lodging (LODG) and yield traits highlighted the influence of plant architecture on productivity in cassava. These findings affirm the applicability of handheld NDVI sensors as cost-effective tools for enhanced phenotyping and selection in cassava breeding programs for rapid genetic gains and varietal development under diverse field conditions.
Detecting leaf diseases is crucial for ensuring crop health and boosting agricultural productivity. An advanced deep learning-based framework is introduced for cassava and groundnut leaf disease detection, incorporating a suite of innovative techniques to enhance classification accuracy. Real-time leaf images are collected from various agricultural environments to capture a wide range of conditions. To improve image quality and segmentation precision, the Contextual Image Enhancement Wiener Filter (CIEWF) is employed for effective noise reduction. Data augmentation is performed using a Generative Adversarial Network (GAN), increasing dataset diversity and improving model generalization. A novel Region of Interest-based Multi-Dimensional Attention Network (ROI-MDAN) is developed to identify and segment critical disease-affected areas within the leaves. For robust feature extraction, the MSFNet-CAM model is proposed, leveraging parallel multi-scale features and incorporating Coordinate Attention to enhance feature fusion and improve classification performance. Furthermore, Gradient-weighted Class Activation Mapping (Grad-CAM) is used to interpret the model's decision-making process by highlighting the influential regions contributing to disease classification. Experimental results validate the effectiveness of the proposed approach, setting a new benchmark for AI-assisted plant disease diagnosis.
Dry matter content (DMC) and starch content (StC) are key quality traits in cassava breeding, yet traditional phenotyping methods are time-consuming and limit scalability. This study aimed to develop and compare predictive models for DMC and StC using near-infrared (NIR) spectroscopy, evaluating two devices-a benchtop spectrometer (Büchi NIRFlex N-500; 1000-2500 nm) and a portable device (QualitySpec Trek; 350-2500 nm)-and assessing the influence of sample type (fresh vs. processed). A total of 3,391 cassava clones from the Embrapa breeding program were analyzed from 2018 to 2023. Reference values were obtained via gravimetric analysis (DMCg), oven drying (DMCo), and manual StC extraction. Spectral data were used to train and validate models using Partial Least Squares (PLS), k-Nearest Neighbors (KNN), and eXtreme Gradient Boosting (XGB). PLS consistently delivered the highest predictive accuracy across traits and devices. KNN slightly outperformed PLS for DMCg using the benchtop device, while XGB was comparable to PLS in select scenarios (e.g., StC with the benchtop: 0.88 vs. 0.89; DMCo with the portable: 0.92 vs. 0.95). Processed samples yielded higher model accuracy than fresh ones. The portable NIR device showed better performance with processed samples and even surpassed the benchtop for DMCg and StC in external validation (0.74 and 0.76 vs. 0.71 and 0.72, respectively). Overall, processed sample preparation significantly improved model performance, and the portable spectrometer proved to be a practical, accurate, and scalable alternative for high-throughput phenotyping in cassava breeding.
Plant diseases establish a serious risk to worldwide food safety and agronomic sustainability, leading to substantial losses in harvest and quality. Precise and appropriate experience of these viruses is necessary for employing effective control procedures, diminishing economic victims and confirming food accessibility. However, current detection methods are often hindered by limitations such as insufficient generalizability across various crops, difficulty in handling complex, and noisy backgrounds, and suboptimal performance when addressing diverse and overlapping disease symptoms. This research is driven by the persistent necessity to tackle these issues and delivers a novel approach to improve disease identification and categorization performance. For this need, propose the Efficient Capsule convolutional Shuffle Attention Network (ECSAN), a comprehensive framework specifically designed to classify diseases across five diverse crops: Apple Leaves, Cassava Leaves, Hibiscus, Hyacinth Bean, and Okra Leaves. The framework integrates a fast gradient-domain weighted guided image filter to denoise and improve image superiority and segmentation is done over a dense swin transformer combined with Unet. Statistical features are extracted using general kernel joint non-negative matrix factorization, and the classification progression is optimized through the Enhanced Osprey Optimization Algorithm (EOOA). Implementation outcomes on five standard datasets prove the superiority of ECSAN, achieving 99.9 % accuracy, 99.98 % recall, 99.97 % precision, 99.98 % F1-score, and 99.99 % specificity. Additionally, ECSAN significantly reduces execution time compared to existing methods, emphasizing its efficiency and applicability in agricultural scenarios. This research underscores the urgency of addressing current detection challenges and establishes a robust foundation for advancements in plant pathology.
Introduction Plant diseases pose a significant threat to global food security and agricultural productivity, making accurate and timely disease identification essential for effective crop management and minimizing economic losses. Although data augmentation techniques such as RICAP improve model robustness, their reliance on randomly extracted image regions can introduce label noise, potentially misleading the training of deep learning models. Methods This study introduces Enhanced-RICAP, an advanced data augmentation technique designed to improve the accuracy of deep learning models for plant disease detection. Enhanced-RICAP replaces random patch selection with an attention module guided by class activation maps, focusing on discriminative regions, Enhanced-RICAP reduces label noise and improves model accuracy for plant disease detection, addressing a key limitation of traditional augmentation methods. The method was evaluated using several deep learning architectures, such as ResNet18, ResNet34, ResNet50, EfficientNet-b, and Xception, on the cassava leaf disease and PlantVillage tomato leaf disease datasets. Results The experimental results demonstrate that Enhanced-RICAP consistently outperforms existing augmentation methods, including CutMix, MixUp, CutOut, Hide-and-Seek, and RICAP, across key evaluation metrics: accuracy, precision, recall, and F1-score. The ResNet18+Enhanced-RICAP configuration achieved 99.86% accuracy on the tomato leaf disease dataset, whereas the Xception+Enhanced-RICAP model attained 96.64% accuracy in classifying four cassava leaf disease categories. Discussion and Conclusion To bridge the gap between research and practical application, the ResNet18+Enhanced-RICAP model was deployed in PlantDisease, a mobile application that enables real-time disease identification and management recommendations. This approach supports sustainable agriculture and strengthens food security by providing farmers with accessible and reliable diagnostic tools.
Global farm productivity is constantly under attack from pests and diseases, resulting in massive crop loss and food insecurity. Manual scouting, expert estimation, and laboratory-based microscopy are time-consuming, prone to human error, and labor-intensive. Although traditional machine learning classifiers such as SVM, Random Forest, and Decision Trees provide better accuracy, they are not field deployable. This article presents a high-performance deep learning fusion model using MobileNetV2 and EfficientNetB0 for real-time detection of pests and diseases in precision farming. The model, trained on the CCMT dataset (24,881 original and 102,976 augmented images in 22 classes of cashew, cassava, maize, and tomato crops), attained a global accuracy of 89.5%, precision and recall of 95.68%, F1-score of 95.67%, and ROC-AUC of 0.95. For supporting deployment in edge environments, methods such as quantization, pruning, and knowledge distillation were employed to decrease inference time to below 10 ms per image. The suggested model is superior to baseline CNN models, including ResNet-50 (81.25%), VGG-16 (83.10%), and other edge lightweight models (83.00%). The optimized model is run on low-power devices such as smartphones, Raspberry Pi, and farm drones without the need for cloud computing, allowing real-time detection in far-off fields. Field trials using drones validated rapid image capture and inference performance. This study delivers a scalable, cost-effective, and accurate early pest and disease detection framework for sustainable agriculture and supporting food security at the global level. The model has been successfully implemented with TensorFlow Lite within Android applications and Raspberry Pi systems.
Abstract The timely and precise identification of diseases in plants is essential for efficient disease control and safeguarding of crops. Manual identification of diseases requires expert knowledge in the field, and finding people with domain knowledge is challenging. To overcome the challenge, computer vision-based machine learning techniques have been proposed by the researchers in recent years. Most of these solutions with the standard convolutional neural network (CNN) approaches use uniform background laboratory setup leaf images to identify the diseases. However, only a few works considered real-field images in their work. Therefore, there is a need for a robust CNN architecture that can identify the diseases in plants in both laboratory and real-field conditioned images. In this paper, we have proposed an Inception-Enhanced Vision Transformer (IEViT) architecture to identify diseases in plants. The proposed IEViT architecture extracts local as well as global features, which improves feature learning. The use of multiple filters with different kernel sizes efficiently uses computing resources to extract relevant features without the need for deeper networks. The robustness of the proposed architecture is established by hyper-parameter tuning and comparison with state-of-the-art. In the experiment, we consider five datasets with both laboratory-conditioned and real-field conditioned images. From the experimental results, we see that the proposed model outperforms state-of-the-art deep learning models with fewer parameters. The proposed model achieves an accuracy rate of 99.23% for the apple leaf dataset, 99.70% for the rice dataset, 97.02% for the ibean dataset, 76.51% for the cassava leaf dataset, and 99.41% for the plantvillage dataset.
Cassava, an essential staple food with diverse applications, has been relatively underexplored in terms of health analysis using vegetation indices. Conventional field surveys face challenges in covering large areas due to resource constraints. Recent advancements in remote monitoring techniques, such as satellite imagery and Unmanned Aerial Vehicles (UAVs), offer a promising alternative. While satellite imagery enables broad-scale surveys, its limited spatial resolution restricts detailed analyses of individual plants or smaller ecosystems. UAV-based vegetation surveys commonly utilize Vegetation Indices (VI) to assess unique spectral information. This study investigated UAV-based methods for mapping cassava distribution in the Telaga Kahuripan smallholder plantation in Bogor, Indonesia, focusing on UAV imagery, segmentation, and vegetation indices to evaluate cassava plant health at 2, 5, and 8 months of age. The results revealed significant variations in vegetation indices across different cassava plant ages. Particularly, the highest values observed at 5 months of age indicated substantial growth, with NDVI and GNDVI values exhibiting R2 ranging from 0.95 to 0.98, indicating a strong correlation. The robust correlation between NDVI and GNDVI implies that both indices can effectively predict plant health using UAV-based monitoring. Comparisons with existing studies suggest potential variations attributable to factors such as geographical location, environmental conditions, and cultivation practices. Understanding these variations is crucial for refining monitoring techniques and informing agricultural practices. Consequently, the findings have implications for enhancing cassava health monitoring and optimizing agricultural practices to ensure sustainable crop production.
Artificial Intelligence (AI) and Deep Learning (DL) are revolutionizing plant disease detection, which is crucial for mitigating crop loss, improving food security, and enhancing yield. Traditional manual methods, such as field visits for disease diagnosis, are labor-intensive, time-consuming, and costly, highlighting the demand for automated, real-time applications. However, plant disease classification faces challenges such as class variations, cluttered backgrounds, lesion scale variations, and the need for robust models. Several Convolutional Neural Network (CNN)-based frameworks have been proposed to tackle these challenges, but they often suffer from limitations such as fixed-size kernels and inefficient feature utilization across deeper layers. These limitations lead to feature loss during the initial stages of feature extraction, reducing the model's overall effectiveness in multi-class plant disease classification. To address these issues, we introduce HFFIncep-Net, a novel Deep Learning framework that fuses hierarchical features with the power of the Inception V3 architecture. The Hierarchical Feature Fusion (HFF) stem extracts multiscale features, which are fused at multiple levels within the network. This enhances feature representation and mitigates information loss in the initial layers. Incorporating two Inception V3 blocks, the model captures a diverse set of features at different scales, further improving classification performance. Additionally, the inclusion of a Global Average Pooling (GAP) layer reduces computational complexity while maintaining high accuracy. To enhance training stability and performance, we employ the Swish activation function, which facilitates smoother gradient propagation during backpropagation. When evaluated on the PlantVillage and Cassava Leaf Disease (CLD) datasets, HFFIncep-Net achieves 94.61% precision, 94.31% recall, 94.31% F1 score, and 97.80% accuracy, outperforming existing CNN-based methods. These results demonstrate that HFFIncep-Net is an effective solution for early-stage, multi-class plant disease classification, with significant potential for practical deployment in agricultural applications.
Cassava ( Manihot esculenta Crantz) is a critical food security crop for sub-Saharan Africa. Efforts to improve cassava through breeding have expanded over the past decade. Crop growth models (CGM) are becoming common place in breeding efforts to expand the inference of evaluations of breeding germplasm to environments that have not been tested and to prepare for breeding for adaptation to future climates. We parameterized a CGM, the CROPGRO-MANIHOT-Cassava model in the DSSAT family of models, using data on 67 clones from the International Institute of Tropical Agriculture cassava breeding program evaluated from 2017 to 2020 and over eight locations in Nigeria using trial and error parameter adjustments and the General Likelihood Uncertainty Estimation method. Our objectives were to assess the feasibility of this large-scale calibration in the context of a cassava breeding program and to identify systematic biases of the model. For each cultivar we calculated the Pearson correlation between model prediction and observation across the environments, as well as root mean squared error and d statistics. As a result of calibration, the correlation coefficient increased from –0.03 to +0.08, the RMSE dropped from 21 t ha -1 to 5 t ha -1 while d increased from 0.23 to 0.44. We found that the model underestimated root yield in dry environments (low precipitation and high temperature) and overestimated root yield in wet environments (high precipitation and low temperature). Our experience suggests both that CGM calibration could become a routine component of the cassava breeding data analysis cycle and that there are opportunities for model improvement.
Early and accurate detection of crop diseases is critical for global food security and efficient agricultural management. Recent advances in deep learning, particularly convolutional neural networks (CNNs) and vision transformers (ViT), have demonstrated exceptional ability to recognize disease symptoms from leaf images. In this article, we present a comprehensive framework for plant disease detection that integrates state-of-the-art deep learning models into smart agriculture systems. We review publicly available datasets (e.g. the PlantVillage dataset with 54,306 leaf images across 14 crop species and 26 disease classes), and discuss data preprocessing and augmentation techniques. We then detail various model architectures: traditional CNNs (e.g. ResNet, MobileNet), efficient CNN variants, ViT-based models, and hybrid CNN–ViT architectures (e.g. FOTCA, AppViT). Our proposed models leverage transfer learning and attention mechanisms to improve accuracy. We describe an experimental setup using multiple leaf-image datasets (tomato, potato, apple, cassava, wheat) and report hypothetical results: for example, our hybrid model achieves ≈99.7% accuracy on PlantVillage and 98–99% on tomato/potato datasets. We include precision, recall, F1 metrics and confusion matrices to analyze performance. Integration into smart farming is discussed: IoT sensors and mobile devices capture leaf images, which are processed by on-device or cloud CNN/ViT models to alert farmers in real time, the depthwise separable convolution block, and the ViT encoding block, respectively. We compare results across models and examine the trade-offs between model complexity and accuracy. Our findings confirm that hybrid CNN–ViT architectures yield the best performance, while lightweight models (e.g. MobileViT, AppViT) enable on-device inference.
Plant diseases are usually a great threat to agriculture in general, for which prompt diagnosis is very important for efficient management action. Most typical leaf diseases that occur on hundreds of plants show their symptoms in the surface leaves. Traditional laboratory diagnosis is costly, time-consuming, requiring a specialized agronomist to confirm results before technology is used. Continually pestering cassava, the most important food crop to about millions in the globe, are leaf diseases that shrink their yields potential. This research narrates a deep learning-based cassava plant disease diagnosis and its classification via leaf image analysis at multi-resolutions. Specifically, it investigates the performance with respect to the four architectures: EfficientNetB4, InceptionV3, ResNet50, and VGG19 - used to differentiate the leaves of cassava ailments. The dense convolutional neural network architecture has been built employing a good amount of plant leaf images from different regions. The dataset features with very rich interclass and intraclass variations along with very complex and challenging scenarios, to which dense neural network adapts pretty well. The performance of the trained model over various parameters validated through five-fold cross-validation as well as unseen data is pretty impressive. With an average cross-validation accuracy of 94 percent with images with complex backgrounds, experimental results indicate that system classifies different plant leaves.
Abstract Crop diseases threaten food security in The Gambia, where agriculture employs 70% of the population. This paper explores the application of computer vision models to help farmers detect and classify crop diseases more effectively, leveraging deep learning techniques—particularly convolutional neural networks (CNNs). Our fine-tuned VGG16 model achieved 91.6% accuracy in identifying diseases like rice blast and cassava mosaic, demonstrating the potential for scalable, low-cost diagnosis. The study evaluates custom CNNs, transfer learning (VGG16, ResNet50, MobileNetV2), and image preprocessing techniques (segmentation, augmentation) to optimize performance for Gambian farm conditions. Beyond technical validation, the paper highlights real-world adoption barriers, including limited technology access, farmer skepticism, and infrastructural gaps. Through surveys and interviews with 120 + farmers, we found that awareness of AI tools strongly correlates with willingness to adopt them (r = 0.63, p
The performance of neural networks is heavily dependent on the integrity of the feature signals. As these signals are extracted and transmitted, they tend to weaken, which can negatively affect their ability to represent and utilize semantic information, particularly in weakly supervised learning tasks. This study aims to address hidden and severely weakened signals. To address the underlying causes, the rectification block of the third stage of PR-ArsenicNetPlus (Positive Rectified ArsenicNetPlus Neural Network) was modified to include a nonlinear fitting method based on the variant Hölder inequality. This method adjusts the magnitude and angular frequency of an extremely weak signal, and its effectiveness is evaluated using Parseval’s relationship. When tested on cassava leaf disease datasets, the proposed method significantly improved the prediction accuracy in 7-fold cross-validation, achieving an accuracy of 96.18 %, a loss of 1.373, and an F1-score of 0.9618. These results outperformed those of ResNet-101, EfficientNet-B5, RepVGG-B3g4, and AlexNet.
Small-scale farmers use cassava as an essential crop for food and nutrition security, owing to its capacity to flourish under adverse situations. In numerous African countries, it serves as a significant source of carbohydrates. Leaf diseases can sometimes damage cassava crops, constraining overall output and farmers' revenue. The ongoing research on cassava disease is fraught with challenges, including a low detection rate, extended processing time, and inadequate precision. This research employs deep transfer learning with Visual Geometry Group (VGG16) models for the diagnosis of Cassava leaf diseases. An experimental study is conducted on a dataset of 5,656 images of cassava, categorized into four distinct disease classifications. Two of the most sophisticated predictions were generated: one identified the healthy leaf, while the other detected the diseases present on the unhealthy leaf. Our suggested deep transfer learning model attains a promising accuracy of 88% and an F1-score of 82% on the public plant disease dataset from the Kaggle repository, achieved through effective hyperparameter fine-tuning. The results of this study strongly advocate for additional research and practical application of deep learning models in plant disease diagnostics.
Abstract The timely and precise identification of diseases in plants is essential for efficient disease control and safeguarding of crops. Manual identification of diseases requires expert knowledge in the field, and finding people with domain knowledge is challenging. To overcome the challenge, computer vision-based machine learning techniques have been proposed by the researchers in recent years. Most of these solutions with the standard convolutional neural network (CNN) approaches use uniform background laboratory setup leaf images to identify the diseases. However, only a few works considered real-field images in their work. Therefore, there is a need for a robust CNN architecture that can identify the diseases in plants in both laboratory and real-field conditioned images. In this paper, we have proposed an Inception-enabled vision transformer (ViT) architecture to identify the diseases in plants. The proposed Inception-enabled ViT architecture extracts local as well as global features, which improves feature learning. The use of multiple filters with different kernel sizes efficiently use computing resources to extract relevant features without the need for deeper networks. The robustness of the proposed architecture is established by hyper-parameter tuning and comparison with state-of-the-art. In the experiment, we consider five datasets with both laboratory-conditioned and real-field conditioned images. From the experimental results, we see that the proposed model outperforms state-of-the-art deep learning models with fewer parameters. The proposed model archives an accuracy rate of 99.17% for the apple leaf dataset, 99.32% for the rice dataset, 96.89% for the ibean dataset, 75.42% for the cassava leaf dataset, and 99.33% for the plantvillage dataset.
The outbreak of cassava diseases poses a serious threat to agricultural economic security and food production systems in tropical regions. Traditional manual monitoring methods are limited by efficiency bottlenecks and insufficient spatial coverage. Although low-altitude drone technology offers advantages such as high resolution and strong timeliness, it faces dual challenges in the field of disease identification, such as complex background interference and irregular disease morphology. To address these issues, this study proposes an intelligent classification method for cassava diseases based on drone imagery and an ED-Swin Transformer. Firstly, we introduced the EMAGE (Efficient Multi-Scale Attention with Grouping and Expansion) module, which integrates the global distribution features and local texture details of diseased leaves in drone imagery through a multi-scale grouped attention mechanism, effectively mitigating the interference of complex background noise on feature extraction. Secondly, the DASPP (Deformable Atrous Spatial Pyramid Pooling) module was designed to use deformable atrous convolution to adaptively match the irregular boundaries of diseased areas, enhancing the model's robustness to morphological variations caused by angles and occlusions in low-altitude drone photography. The results show that the ED-Swin Transformer model achieved excellent performance across five evaluation metrics, with scores of 94.32%, 94.56%, 98.56%, 89.22%, and 96.52%, representing improvements of 1.28%, 2.32%, 0.38%, 3.12%, and 1.4%, respectively. These experiments demonstrate the superior performance of the ED-Swin Transformer model in cassava classification networks.
Cassava is a tuberous edible plant native to the American tropics and is essential for its versatile applications including cassava flour, bread, tapioca, and laundry starch. Cassava leaf diseases reduce crop yields, elevate production costs, and disrupt market stability. This places significant burdens on farmers and economies while highlighting the need for effective management strategies. Traditional methods of manual disease diagnosis are costly, labor-intensive, and time-consuming. This research aims to address the challenge of accurate disease classification by overcoming the limitations of existing methods, which encounter difficulties with the complexity and variability of leaf disease symptoms. To the best of our knowledge, this is the first study to propose a novel dual-track feature aggregation architecture that integrates the Residual Inception Positional Encoding Attention (RIPEA) Network with EfficientNet for the classification of cassava leaf diseases. The proposed model employs a dual-track feature aggregation architecture which integrates the RIPEA Network with EfficientNet. The RIPEA track extracts significant features by leveraging residual connections for preserving gradients and uses multi-scale feature fusion for combining fine-grained details with broader patterns. It also incorporates Coordinate and Mixed Attention mechanisms which focus on cross-channel and long-range dependencies. The extracted features from both tracks are aggregated for classification. Furthermore, it incorporates an image augmentation method and a cosine decay learning rate schedule to improve model training. This improves the ability of the model to accurately differentiate between Cassava Bacterial Blight (CBB), Brown Streak Disease (CBSD), Green Mottle (CGM), Mosaic Disease (CMD), and healthy leaves, addressing both local textures and global structures. Additionally, to enhance the interpretability of the model, we apply Grad-CAM to provide visual explanations for the model's decision-making process, helping to understand which regions of the leaf images contribute to the classification results. The proposed network achieved a classification accuracy of 93.06%.
Cassava is a vital crop for millions of farmers worldwide, but its cultivation is threatened by various destructive diseases. Current detection methods for cassava diseases are costly, time-consuming, and often limited to controlled environments, making them unsuitable for large-scale agricultural use. This study aims to develop a deep learning framework that enables early, accurate, and efficient detection of cassava diseases in real-world conditions. We propose a self-supervised object segmentation technique, combined with a progressive learning algorithm (PLA) that incorporates both triplet loss and classification loss to learn robust feature embeddings. Our approach achieves superior performance on the Cassava Leaf Disease Classification (CLDC) dataset from the Kaggle competition, with an accuracy of 91.43%, outperforming all other participants. The proposed method offers a practical and efficient solution for cassava disease detection, demonstrating the potential for large-scale, real-world application in agriculture.
Abstract Detecting diseases in plant leaves is very crucial for making sure that there is enough food worldwide and that we can continue with sustainable farming. Plant diseases are big threats to the output of crops, safety in food and economic security worldwide. The importance of finding and controlling these diseases quickly is very high to lessen their effect and maintain a good balance in farming methods. The main reason why this study started was because of how important it is to have precise and dependable methods for detecting diseases in plant leaves within the field of agriculture. The development of AI-driven Smart Detection for Plant Leaves (AI-SCAN) is a pliable and robust technique that will raise the reliability and accuracy of plant leaf disease diagnosis. The construction procedure of the recommended model highlights the fundamental technique employed in our proposed study. Here, the Transcendental Residual Convolutional Swin Transformer (TRCST) technique – which was invented lately and is exclusive, has been used to accurately identify the type of plant disease from the given images. Furthermore, Nomadic People’s most recent hybrid optimization model incorporates photon optimizer (NP2O) to fine-tune TRCST parameters, significantly improving AI-SCAN’s overall disease detection performance. The merging of TRCST with NP2O into AI-SCAN gives an original way that can be used for finding plant leaf disease, providing better trustworthiness, precision and speed when diagnosing sicknesses on plants. This method deals with the main issues linked to agriculture’s sustained capacity for producing food and defending crops against destructive diseases. The Plant Village, Cassava, and Jasmine datasets are three different publicly available, large-scale plant leaf datasets that were used to examine the results and implications of the proposed AI-SCAN.
Cassava Mosaic Disease (CMD) is a severe viral infection affecting cassava crops, spread by whiteflies and infected plant material, causing significant yield losses globally. Traditional CMD detection methods are labor-intensive and slow. This study developed a data-driven predictive model using remote sensing and climatic data to forecast CMD onset in cassava crops. We applied a derivative approach to identify local minima at the cassava growing season’s onset, referred to as the Start of Season (SOS), by differentiating the Normalized Difference Vegetation Index (NDVI) time series data from Sentinel-2 L2A satellite imagery. When compared to conventionally estimated SOS, this method achieved a Root Mean Square Error (RMSE) of 3 days. Climatic data, aggregated over various temporal periods before and after the SOS, were used as predictors for CMD onset through a multistage predictive modeling approach. This approach incorporated ensemble variable selection using five methods: extreme gradient boosting, random forest, support vector machine, artificial neural network, and logistic regression to identify the most relevant climatic variables. Additionally, we employed a CART regression tree approach to estimate the most predictive climatic features associated with CMD onset. Findings indicated that relative humidity conditions of ≤ 70.65 % for at least four consecutive days within the first ten days after SOS were crucial for maximizing the accuracy of CMD onset date predictions, achieving an average ensemble prediction error of 17.7 days (measured by RMSE). The extreme gradient boosting algorithm outperformed the other models, achieving the highest accuracy in predicting CMD onset dates, with a success rate of 61.5 % and an RMSE of 8 days. Our findings indicate that integrating climatic factors, especially by establishing a timeframe based on the remote sensing-derived SOS date, is essential for identifying key CMD triggering factors and disease onset, applicable at both the farm scale and across broader satellite-covered regions.
Detecting cassava leaf disease is challenging because it is hard to identify diseases accurately through visual inspection. Even trained agricultural experts may struggle to diagnose the disease correctly which leads to potential misjudgements. Traditional methods to diagnose these diseases are time-consuming, prone to error, and require expert knowledge, making automated solutions highly preferred. This paper explores the application of advanced deep learning techniques to detect as well as classify cassava leaf diseases which includes EfficientNet models, DenseNet169, Xception, MobileNetV2, ResNet models, Vgg19, InceptionV3, and InceptionResNetV2. A dataset consisting of around 36,000 labelled images of cassava leaves, afflicted by diseases such as Cassava Brown Streak Disease, Cassava Mosaic Disease, Cassava Green Mottle, Cassava Bacterial Blight, and healthy leaves, was used to train these models. Further the images were pre-processed by converting them into grayscale, reducing noise using Gaussian filter, obtaining the region of interest using Otsu binarization, Distance transformation, as well as Watershed technique followed by employing contour-based feature selection to enhance model performance. Models, after fine-tuned with ADAM optimizer computed that among the tested models, the hybrid model (DenseNet169 + EfficientNetB0) had superior performance with classification accuracy of 89.94% while as EfficientNetB0 had the highest values of precision, recall, and F1score with 0.78 each. The novelty of the hybrid model lies in its ability to combine DenseNet169's feature reuse capability with EfficientNetB0's computational efficiency, resulting in improved accuracy and scalability. These results highlight the potential of deep learning for accurate and scalable cassava leaf disease diagnosis, laying the foundation for automated plant disease monitoring systems.
Large-scale phenotyping using unmanned aerial vehicles (UAVs) has been considered an important tool for plant selection. This study aimed to estimate the correlations between agronomic data and vegetation indices (VIs) obtained at different flight heights and to select prediction models to evaluate the potential use of aerial imaging in cassava breeding programs. Various VIs were obtained and analyzed using mixed models to derive the best linear unbiased predictors, heritability parameters, and correlations with various agronomic traits. The VIs were also used to build prediction models for agronomic traits. Aerial imaging showed high potential for estimating plant height, regardless of flight height (r = 0.99), although lower-altitude flights (20 m) resulted in less biased estimates of this trait. Multispectral sensors showed higher correlations compared to RGB, especially for vigor, shoot yield, and fresh root yield (−0.40 ≤ r ≤ 0.50). The heritability of VIs at different flight heights ranged from moderate to high (0.51 ≤ HCullis2 ≤ 0.94), regardless of the sensor used. The best prediction models were observed for the traits of plant vigor and dry matter content, using the Generalized Linear Model with Stepwise Feature Selection (GLMSS) and the K-Nearest Neighbor (KNN) model. The predictive ability for dry matter content increased with flight height for the GLMSS model (R2 = 0.26 at 20 m and R2 = 0.44 at 60 m), while plant vigor ranged from R2 = 0.50 at 20 m to R2 = 0.47 at 40 m in the KNN model. Our results indicate the practical potential of implementing high-throughput phenotyping via aerial imaging for rapid and efficient selection in breeding programs.
Cassava (Manihot esculenta Crantz), a crucial food source for millions, has experienced significant crop yield losses due to various plant diseases. To address this issue, extensive research has been conducted on the Cassava 2020 dataset released by the Makerere AI Lab. Images in this dataset are captured in complex real-world environments, adding difficulty to the classification process. While several deep learning models have shown promising results, we believe that reducing interference from overlapping leaves and complex backgrounds could further enhance classification performance. To accomplish this, up to five leaves of interest have been identified and annotated in each image. These annotated leaves are then used to train a detection model: Faster R-CNN with ResNet-101, designed to automatically detect leaves of interest in the images. The identified leaves are separated from the background through a masking process, producing masked images (Mi), where i indicates the maximum number of leaves in the masked images, with i ranging from 1 to 5. The masked images M1 to M5 are then trained on various CNN models; i.e. EfficientNetB1, DenseNet121, ResNet50, and Xception, for Cassava disease classification. The results show that classifiers trained with M3 improve the accuracy by 2.13% to 3.06% compared to models trained with original images. The main contribution of this work is a novel technique capable of identifying leaves of interest from complex backgrounds to produce masked images, which improve the accuracy of various CNN-based classification models, and a novel leaf annotation where up to 5 leaves of interest have been annotated for each image in Cassava 2020 dataset. Annotated images can serve as a valuable resource for advanced cassava disease analysis, detection, and classification.
The accurate and fast assessment of plant diseases and fruits is important for sustainable and productive agriculture. However, typically manual methods are adopted for the assessment of plant diseases and fruit, which is a time-consuming, resource-intensive, and error-prone approach. A few artificial intelligence (AI)-based methods have been introduced to automate this process, but they show limitations in attaining high performance in challenging imaging conditions such as occlusions, poor lighting, noise, indistinctive boundaries, and excessive morphological variations. Moreover, current approaches also lack in providing computationally efficient solutions. To overcome these issues, two novel architectures are developed to segment plant diseases and fruits with higher segmentation performance and less computational requirements. The efficient feature fusion segmentation network (EFFS-Net) is the base network, and a multi-scale dilated feature fusion segmentation network (MDFS-Net) is the final network of this study. EFFS-Net uses an identity skip path-based feature fusion mechanism with an efficient grouped convolutional depth (EGCD) to provide satisfactory segmentation performance with high computational efficiency. MDFS-Net uses a multi-scale feature fusion mechanism and fuses low-level information in the EGCD and other sections of the network to learn detailed input features for delivering promising performance even in challenging imaging conditions. MDFS-Net also applies multiple receptive fields to the low-level information in the effective receptive field processing (ERFP) block for fusion near the pixel classification stage for further performance enhancement. Both networks are evaluated using a Brazilian Arabica coffee leaf (BRACOL) image dataset, an Australian Center for Field Robotics orchard fruit (apple) dataset, and a necrotized cassava root cross-section image dataset. The proposed method provides a promising segmentation performance, achieving dice similarity coefficients of 88.81 %, 95.01 %, and 86.04 % with performance improvement of 3.5 %, 2.6 %, and 1.07 % on the three datasets, respectively, with approximately ten times less number of required trainable parameters compared with the state-of-the-art methods. • Architectures (EFFS-Net and MDFS-Net) for plant disease and fruit segmentation. • EFFS-Net uses efficient grouped convolutional depth. • MDFS-Net uses multi-scale and multi-receptive field-based feature fusion mechanisms. • Our trained models and codes are publicly available via Github site.
Abstract Growth rate of crops is significantly lowered by illnesses that affect plants. It is impossible for anyone to eat the crops since they are tainted with various diseases. Farmers may suffer enormous losses as a consequence. Since cassava is an important food source in several countries, the financial system might be seriously damaged by the issue at hand. Traditional plant pathogen detection is labour-intensive and error-prone. It is not typically a dependable strategy to identify and stop the spread of plant viruses. Innovative technologies like deep learning as well as machine learning might aid in the early detection of plant diseases as an approach to get around these problems. The main goal of the work is to employ deep learning to image classification in order to accurately identify diseases that especially impact cassava plants. This recognition may make it possible to implement preventative measures like the specific application of chemical pesticides or confinement of contaminated crops. Each and every training and testing image comes from a rural area in the natural world. Using a specific collection of information, the simulation has been verified to ascertain its true results. The installation of a precise disease identification and mitigation model has the potential to significantly increase the durability of the cassava crop, improving food production and the quality of life for millions of people who are dependent upon this valuable crop.
Background Post-harvest physiological deterioration (PPD) poses a significant challenge to the cassava industry, leading to substantial economic losses. This study aims to address this issue by developing a comprehensive framework in collaboration with cassava breeders. Results Advanced deep learning (DL) techniques such as Segment Anything Model (SAM) and YOLO foundation models (YOLOv7, YOLOv8, YOLOv9, and YOLO-NAS), were used to accurately categorize PPD severity from RGB images captured by cameras or cell phones. YOLOv8 achieved the highest overall mean Average Precision (mAP) of 80.4%, demonstrating superior performance in detecting and classifying different PPD levels across all three models. Although YOLO-NAS had some instability during training, it demonstrated stronger performance in detecting the PPD_0 class, with a mAP of 91.3%. YOLOv7 exhibited the lowest performance across all classes, with an overall mAP of 75.5%. Despite challenges with similar color intensities in the image data, the combination of SAM, image processing techniques such as RGB color filtering, and machine learning (ML) algorithms was effective in removing yellow and gray color sections, significantly reducing the Mean Absolute Error (MAE) in PPD estimation from 20.01 to 15.50. Moreover, Artificial Intelligence (AI)-based algorithms allow for efficient analysis of large datasets, enabling rapid screening of cassava roots for PPD symptoms. This approach is much faster and more streamlined compared to the labor-intensive and time-consuming manual visual scoring methods. Conclusion These results highlight the significant advancements in PPD detection and quantification in cassava samples using cutting-edge AI techniques. The integration of YOLO foundation models, alongside SAM and image processing methods, has demonstrated promising precision even in scenarios where experts struggle to differentiate closely related classes. This AI-powered model not only effectively streamlines the PPD assessment in the pre-breeding pipeline but also enhances the overall effectiveness of cassava breeding programs, facilitating the selection of PPD-resistant varieties through controlled screening. By improving the precision of PPD assessments, this research contributes to the broader goal of enhancing cassava productivity, quality, and resilience, ultimately supporting global food security efforts.
Cassava ( Manihot esculenta Crantz) is a critical food security crop for sub-Saharan Africa. Efforts to improve cassava through breeding have expanded over the past decade. At the same time, crop growth models (CGM) are becoming common place in breeding efforts to expand the inference of evaluations of breeding germplasm to environments that have not been tested and to prepare for breeding for adaptation to future climates. We parameterized a CGM, the CROPGRO-MANIHOT-Cassava model in the DSSAT family of models, using data on 67 clones from the International Institute of Tropical Agriculture cassava breeding program evaluated from 2017 to 2020 and over eight locations in Nigeria using trial and error parameter adjustments and the General Likelihood Uncertainty Estimation method. Our objectives were to assess the feasibility of this large-scale calibration in the context of a cassava breeding program and to identify systematic biases of the model. For each cultivar we calculated the Pearson correlation between model prediction and observation across the environments, as well as root mean squared error and d statistics. As a result of calibration, the correlation coefficient increased from −0.03 to +0.08, the RMSE dropped from 21 t ha -1 to 5 t ha -1 while d increased from 0.23 to 0.44. We found that the model underestimated root yield in dry environments (low precipitation and high temperature) and overestimated root yield in wet environments (high precipitation and low temperature). Our experience suggests both that CGM calibration could become a routine component of the cassava breeding data analysis cycle and that there are opportunities for model improvement.
Leaf diseases have become more prevalent in recent years due to climate change, increased growth of outdoor air pollutants, and global warming. They may severely damage crop yield, leading to detrimental effects on global food security. The timely and precise detection of leaf diseases is thus crucial for preventing their spread and ensuring the sustainability of agricultural production. In this paper, we introduce a transformer-based leaf disease detection model, namely FormerLeaf along with two optimization methods to enhance the model performance. In more detail, we propose the Least Important Attention Pruning (LeIAP) algorithm to select the most important attention heads of each layer in the Transformer model. It could reduce the model size up to 28% and accelerate the evaluation speed by 15% with about 3% accuracy enhancement. In addition, we employ the sparse matrix-matrix multiplication (SPMM) to calculate matrix correlation in the model. This reduces the model’s complexity from O(n2) to O(n2p), resulting in lowering training time by 10% while keeping a similar performance. The evaluation results on the Cassava Leaf Disease Dataset show that our proposal outperforms the state-of-the-art models in most cases.
The prompt detection of plant diseases mitigates adverse effects on plants. Convolutional neural networks (CNN) and intense learning are extensively utilized in computer vision and recognition of pattern tasks. Scientists presented several DL algorithms for the detection of plant illnesses. Deep learning (DL) models need many parameters, resulting in extended training durations and complicated implementation on compact devices. This research presents a unique DL model utilizing the inception tier and residual connections. Depthwise differentiated convolution is employed to decrease the variable count. The suggested model has undergone training and evaluation using three distinct plant disease databases. The level of accuracy achieved on the PlantVillage database is 97.2%, on the rice disease database is 98.4%, and on the cassava database is 96.3%. The suggested model attains superior accuracy relative to state-of-the-art DL methods while utilizing fewer variables.
Background Cassava is a high-carbohydrate crop that is at risk of viral infections. The production rate and quality of cassava crops are affected by several diseases. However, the manual identification of diseases is challenging and requires considerable time because of the lack of field professionals and the limited availability of clear and distinct information. Consequently, the agricultural management system is seeking an efficient and lightweight method that can be deployable to edged devices for detecting diseases at an early stage. To address these issues and accurately categorize different diseases, a very effective and lightweight framework called CDDNet has been introduced. We used MobileNetV3Small framework as a backbone feature for extracting optimized, discriminating, and distinct features. These features are empirically validated at the early intermediate stage. Additionally, we modified the soft attention module to effectively prioritize the diseased regions and enhance significant cassava plant disease-related features for efficient cassava disease detection. Results Our proposed method achieved accuracies of 98.95%, 97.03%, and 98.25% on Cassava Image Dataset, Cassava Plant Disease Merged (2019-2020) Dataset, and the newly created Cassava Plant Composite Dataset, respectively. Furthermore, the proposed technique outperforms previous state-of-the-art methods in terms of accuracy, parameter count, and frames per second values, ultimately making the proposed CDDNet the best one for real-time processing. Conclusion Our findings underscore the importance of a lightweight and efficient technique for cassava disease detection and classification in a real-time environment. Furthermore, we highlight the impact of modified soft attention on model performance. © 2024 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Cassava is the third largest source of carbohydrates for human consumption worldwide; however, it is highly susceptible to viral and bacterial diseases, which pose a significant threat to food security. The advancement of deep learning algorithms in precision agriculture holds the key to enabling the early classification of plant diseases, thereby leading to enhanced crop yields and ultimately stabilizing food security. In the coarse-grained label discrimination task of weak supervision learning, high-quality semantic features contain abundant semantic description information, which plays a crucial role in constructing a precise description of plant disease discrimination in tanglesome field circumstances and directly influences the performance of neural networks. Thus, a multiattention IBN anti-aliasing neural network (MAIANet) was proposed to improve the classification accuracy of cassava leaf disease classification by improving the feature quality in the coarseness label classification task. The proposed MAIANet neural network includes two innovative approaches. First, the multiattention method was designed to scale the feature signals twice to adjust the angular frequency of the feature signals in the residual branch for optimal feature fitting within the residual unit. Second, the anti-aliasing block extracts the high-frequency component feature and optimizes the quantization result of the pooling operation to depress the aliasing signal in the down-sampled feature maps. When the proposed method was tested and validated on the cassava dataset, the results showed that the prediction accuracy of the proposed method significantly improved, with an accuracy of 95.83 %, a loss of 1.720, and an F1-score of 0.9585, outperforming V2-ResNet-101, EfficientNet-B5, RepVGG-B3g4, and AlexNet with significant margins. Based on the above experimental results, the proposed algorithm is suitable for classifying cassava leaf diseases.
Abstract Global warming and lack of immunity in crops have recently resulted in a significant increase in the spread of agricultural diseases. This leads to large-scale crop destruction, less cultivation, and ultimately financial loss for farmers. Identification and treatment of illnesses have become a big issue because of the fast development in disease diversity and lack of farmer knowledge. This paper investigates the application of deep learning for crop disease prediction using a newly acquired dataset of leaf images from Ghana. The dataset focuses on four major crops: cashew, tomato, cassava, and maize. The paper introduces hybrid deep learning models in terms of various evaluation metrics in identifying healthy and diseased plants based on leaf images. This paper also developed a novel hybrid model for this new dataset. The hybrid model ResNet50 + VGG16 resulted in higher precision and accuracy in its predictions, evidencing strong performance and reliability. This work contributes to the development of accurate and accessible tools for crop disease diagnosis, potentially leading to improved agricultural practices and increased crop yields. Through the integration of newer and advanced deep learning techniques, this research will provide a significant step in the field of agriculture for monitoring crop health disease and prediction.
Food security depends on the early detection of agricultural diseases, particularly in Sub-Saharan Africa. Professionals visually evaluate the plants by searching for disease indications on the leaves to diagnose cassava infections, a notoriously subjective process. Farmers in remote areas may be able to monitor their crops without the assistance of specialists if crop diseases are automatically detected and classified. This could aid in the more accurate diagnosis of diseases by professionals. Crop disease classification and early identification have benefited from the application of machine learning techniques. Despite their excellent accuracy, there is no single machine learning model that can provide optimal results on all the datasets. In this study, a Bayesian optimization of the Convolutional Neural Network (CNN) model was proposed. The model was trained using the spectral dataset. Since spectral data is highly dimensional, dimensionality reduction was performed on the dataset using PCA. The experimental results revealed that the proposed model had an accuracy of 85.19%, Precision of 85.23%, Recall of 85.16%, and F1 score of 85.20%. Also, the proposed model had an AUC of 0.95 which demonstrates excellent performance. However, there is still a need to improve the overall performance of the proposed model and we recommend the use of a pretrained transfer learning approach in future studies.
Cassava is a most important carbohydrate human food consumed in many African and Asian countries. Cassava leaf disease is the major issue which affects production. Automatic early cassava leaf disease detection through deep learning models and transfer learning models were used for multiclass classification with different approaches. Existing approaches deal with imbalanced dataset for predicting the classes. This research work develops an approach based on hybrid Ensemble - deep transfer model approach for early leaf disease detection. Data augmentation was applied to the raw data for balancing the dataset. Three distinct new hybrid models namely Ensemble(InceptionV3+DenseNet-BC-121-32 + Xception), Ensemble(ResNet50V2+DenseNet-BC-121-32), Ensemble(ResNet50V2+ResNet50) were developed. The proposed model shows high performance results. A broad comparison of the proposed model was performed with custom based Convolutional Neural Network and pre-trained models. Highest accuracy of 88.83% and 97.89% was obtained in ensemble based approach that combined InceptionV3, Xception, DenseNet-BC-121-32 for five class and two class classification respectively.
Abstract Climate change-resilient crops like cassava are projected to play a key role in 21st-century food security. However, cassava production in East Africa is limited by RNA viruses that cause cassava brown streak disease (CBSD). CBSD typically causes subtle or no symptoms on stems and leaves, while destroying the root tissue, which means farmers are often unaware their fields are infected until they have a failed harvest. The subtle symptoms of CBSD have made it difficult to study the spread of the disease in fields. We will use an engineering advancement, our active multispectral imager (MSI), to rapidly determine the infection status of plants in the field in Tanzania. The MSI observes leaves using many different wavelengths, and the resulting light spectra are interpreted by machine learning models trained on cassava leaf scans. Under laboratory conditions, the MSI detects CBSD infection with 95% accuracy at 28 days post-infection, when plants have no visible symptoms. Our multinational team is studying and modeling the spread of CBSD to assess the efficacy of using the MSI to detect and remove infected cassava plants from fields before CBSD can spread. In addition to improving the food security of people who eat cassava in sub-Saharan Africa, our technology and modeling framework may be useful in diseases of other vegetatively propagated crops such as banana/plantain, potato, sweet potato, and yam. Information © The Authors 2024
Abstract Determination of pasting properties of high quality cassava flour using rapid visco analyzer is expensive and time consuming. The use of mobile near infrared spectroscopy (SCiO™) is an alternative high throughput phenotyping technology for predicting pasting properties of high quality cassava flour traits. However, model development and validation are necessary to verify that reasonable expectations are established for the accuracy of a prediction model. In the context of an ongoing breeding effort, we investigated the use of an inexpensive, portable spectrometer that only records a portion (740–1070 nm) of the whole NIR spectrum to predict cassava pasting properties. Three machine-learning models, namely glmnet, lm, and gbm, implemented in the Caret package in R statistical program, were solely evaluated. Based on calibration statistics (R 2 , RMSE and MAE), we found that model calibrations using glmnet provided the best model for breakdown viscosity, peak viscosity and pasting temperature. The glmnet model using the first derivative, peak viscosity had calibration and validation accuracy of R 2 = 0.56 and R 2 = 0.51 respectively while breakdown had calibration and validation accuracy of R 2 = 0.66 and R 2 = 0.66 respectively. We also found out that stacking of pre-treatments with Moving Average, Savitzky Golay, First Derivative, Second derivative and Standard Normal variate using glmnet model resulted in calibration and validation accuracy of R 2 = 0.65 and R 2 = 0.64 respectively for pasting temperature. The developed calibration model predicted the pasting properties of HQCF with sufficient accuracy for screening purposes. Therefore, SCiO™ can be reliably deployed in screening early-generation breeding materials for pasting properties.
Cooking time is a crucial determinant of culinary quality of cassava roots and incorporating it into the early stages of breeding selection is vital for breeders. This study aimed to assess the potential of near-infrared spectroscopy (NIRS) in classifying cassava genotypes based on their cooking times. Five cooking times (15, 20, 25, 30, and 40 minutes) were assessed and 888 genotypes evaluated over three crop seasons (2019/2020, 2020/2021, and 2021/2022). Fifteen roots from five plants per plot, featuring diameters ranging from 4 to 7 cm, were randomly chosen for cooking analysis and spectral data collection. Two root samples (15 slices each) per genotype were collected, with the first set aside for spectral data collection, processed, and placed in two petri dishes, while the second set was utilized for cooking assessment. Cooking data were classified into binary and multiclass variables (CT4C and CT6C). Two NIRs devices, the portable QualitySpec® Trek (QST) and the benchtop NIRFlex N-500 were used to collect spectral data. Classification of genotypes was carried out using the K-nearest neighbor algorithm (KNN) and partial least squares (PLS) models. The spectral data were split into a training set (80%) and an external validation set (20%). For binary variables, the classification accuracy for cassava cooking time was notably high ( RCal2 ranging from 0.72 to 0.99). Regarding multiclass variables, accuracy remained consistent across classes, models, and NIR instruments (~0.63). However, the KNN model demonstrated slightly superior accuracy in classifying all cooking time classes, except for the CT4C variable (QST) in the NoCook and 25 min classes. Despite the increased complexity associated with binary classification, it remained more efficient, offering higher classification accuracy for samples and facilitating the selection of the most relevant time or variables, such as cooking time ≤ 30 minutes. The accuracy of the optimal scenario for classifying samples with a cooking time of 30 minutes reached RCal2 = 0.86 and RVal2 = 0.84, with a Kappa value of 0.53. Overall, the models exhibited a robust fit for all cooking times, showcasing the significant potential of NIRs as a high-throughput phenotyping tool for classifying cassava genotypes based on cooking time.
BACKGROUND: The purpose of this study was to investigate the potential of hyperspectral imaging for the characterization of cooking quality parameters, dry matter content (DMC), water absorption (WAB), and texture in cassava genotypes contrasting for their cooking quality. RESULTS: Hyperspectral images were acquired on cooked and fresh intact longitudinal and transversal slices from 31 cassava genotypes harvested in March 2022 in Colombia. Different chemometric methods were tested for the quantification of DMC, WAB, and texture parameters. Data analysis was conducted through partial least squares regression, K nearest neighbors regression, support vector machine regression and CovSel multiple linear regression (CovSel_MLR). Efficient performances were obtained for DMC using CovSel_MLR with, coefficient of multiple determination Rp2=0.94, root‐mean‐square error of prediction RMSEP = 0.96 g/100 g, and ratio of the standard deviation values RPD = 3.60. High heterogeneity was observed between contrasting genotypes. The predicted distribution of DMC within the root can be homogeneous or heterogeneous depending on the genotype. Weak predictions were obtained for WAB and texture parameters. CONCLUSIONS: This study showed that hyperspectral imaging could be used as a high‐throughput phenotyping tool for the visualization of DMC in contrasting cooking quality genotypes. Further improvement of protocols and larger datasets are required for WAB and texture quality traits. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
BACKGROUND: Cassava retting ability and the textural qualities of cooked fufu are important quality traits. Cassava retting is a complex process in which soaking causes tissue breakdown, starch release, and softening. The rate at which various traits linked to it evolve varies greatly during fufu processing. According to the literature, there is no standard approach for determining retting ability. The retting indices and textural properties of fufu were measured using both manual and instrumental approaches. RESULTS: Different protocols were developed to classify 64 and 11 cassava genotypes into various groups based on retting ability and textural qualities, respectively. The retting protocols revealed considerable genetic dissimilarities in genotype classification: foaming ability and water clarity should be measured at 24 h, while penetrometer, hardness, turbidity, pH, and total titratable acidity data are best collected after 36 h. The stepwise regression model revealed that pH, foaming ability, and dry matter content are the best multivariates (with the highest R²) for predicting cassava retting. These predictors were used to develop an index for assessing the retting ability of cassava genotypes. The retting index developed showed a significant relationship with dry matter content and fufu yield. The study also showed significant correlations between instrumental cohesiveness and sensory smoothness (r = −0.75), moldability (r = −0.62), and stretchability (r = 0.78). Instrumental cohesiveness can correctly estimate fufu smoothness (R² = 0.56, P = 0.008) and stretchability (R² = 0.60, P = 0.005). CONCLUSION: pH, foaming ability, and dry matter content are the best traits for predicting cassava retting ability, while instrumental cohesiveness can effectively estimate fufu smoothness and stretchability. © 2024 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Abstract Background The sweet potato whitefly ( Bemisia tabaci ) is a globally important insect pest that damages crops through direct feeding and by transmitting viruses. Current B. tabaci management revolves around the use of insecticides, which are economically and environmentally costly. Host plant resistance is a sustainable option to reduce the impact of whiteflies, but progress in deploying resistance in crops has been slow. A major obstacle is the high cost and low throughput of screening plants for B. tabaci resistance. Oviposition rate is a popular metric for host plant resistance to B. tabaci because it does not require tracking insect development through the entire life cycle, but accurate quantification is still limited by difficulties in observing B. tabaci eggs, which are microscopic and translucent. The goal of our study was to improve quantification of B. tabaci eggs on several important crop species: cassava, cowpea, melon, sweet potato and tomato. Results We tested a selective staining process originally developed for leafhopper eggs: submerging the leaves in McBryde’s stain (acetic acid, ethanol, 0.2% aqueous acid Fuchsin, water; 20:19:2:1) for three days, followed by clearing under heat and pressure for 15 min in clearing solution (LGW; lactic acid, glycerol, water; 17:20:23). With a less experienced individual counting the eggs, B. tabaci egg counts increased after staining across all five crops. With a more experienced counter, egg counts increased after staining on melons, tomatoes, and cowpeas. For all five crops, there was significantly greater agreement on egg counts across the two counting individuals after the staining process. The staining method worked particularly well on melon, where egg counts universally increased after staining for both counting individuals. Conclusions Selective staining aids visualization of B. tabaci eggs across multiple crop plants, particularly species where leaf morphological features obscure eggs, such as melons and tomatoes. This method is broadly applicable to research questions requiring accurate quantification of B. tabaci eggs, including phenotyping for B. tabaci resistance.
Background Cassava retting ability and the textural qualities of cooked fufu are important quality traits. Cassava retting is a complex process in which soaking causes tissue breakdown, starch release, and softening. The rate at which various traits linked to it evolve varies greatly during fufu processing. According to the literature, there is no standard approach for determining retting ability. The retting indices and textural properties of fufu were measured using both manual and instrumental approaches. Results Different protocols were developed to classify 64 and 11 cassava genotypes into various groups based on retting ability and textural qualities, respectively. The retting protocols revealed considerable genetic dissimilarities in genotype classification: foaming ability and water clarity should be measured at 24 h, while penetrometer, hardness, turbidity, pH, and total titratable acidity data are best collected after 36 h. The stepwise regression model revealed that pH, foaming ability, and dry matter content are the best multivariates (with the highest R 2 ) for predicting cassava retting. These predictors were used to develop an index for assessing the retting ability of cassava genotypes. The retting index developed showed a significant relationship with dry matter content and fufu yield. The study also showed significant correlations between instrumental cohesiveness and sensory smoothness (r = -0.75), moldability (r = -0.62), and stretchability (r = 0.78). Instrumental cohesiveness can correctly estimate fufu smoothness (R 2 = 0.56, P = 0.008) and stretchability (R 2 = 0.60, P = 0.005). Conclusion pH, foaming ability, and dry matter content are the best traits for predicting cassava retting ability, while instrumental cohesiveness can effectively estimate fufu smoothness and stretchability. © 2024 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Background Cassava leaf disease detection is a major problem since it is very difficult to identify the disease in naked-eye observation and even experts such as agricultural scientists can fail in this task. The approach we use in this paper has the potential to overcome this problem. Materials and Methods In this, we propose an approach based on artificial intelligence for leaf disease detection using deep learning with generative adversarial networks (GAN). Our experimental study used a dataset including 12880 cassava leaf pictures generated using CycleGAN showing five major disease classes. In order to avoid overfitting, a GAN architecture is proposed for data augmentation using two networks, i.e ., a Generator and a Discriminator. The generator is trained to generate similar data samples as the original data Results The proposed approach achieved an accuracy of 99.51% for the classification of healthy or unhealthy leaf images, which outperformed existing methods. Discussion The discriminator is trained to distinguish between the unique and generated sample records, as actual or fake. To classify cassava images into five categories of diseases, a combination of machine learning models has been trained on original and generated images. The proposed approach showed better accuracy compared to the existing methods. Conclusion The proposed deep learning-based method can be used as a tool for early disease diagnosis in cassava leaf disease detection and classification
With artificial intelligence and deep learning development, crop disease recognition methods leverage deep networks for automatic feature learning but rely on the volume of training data. Addressing the scarcity of data in agriculture, few-shot learning (FSL) and multi-modal learning have become focal points. However, existing methods are confined to a single modality or insufficiently exploit cross-modal features. To address this, we propose a multi-modal contrastive learning approach integrating images and text to tackle the problem of small-sample recognition. This method combines CLIP multi-modal pre-training with cross-attention, termed ITIMCA. Experimental validation demonstrates the effectiveness of our approach in cassava leaf disease recognition tasks under natural conditions. Experimental results show that the proposed model achieved an accuracy of 78.00%, a precision of 88.48%, a recall rate of 80.00%, and an F1-Score of 79.00% on the cassava leaf disease identification and classification dataset. These results suggest that the proposed network effectively identifies cassava leaf diseases.
In this study, near–infrared spectroscopy (NIRS) was used to predict quality of cassava leaves and roots based on their own spectra and also to predict root quality based on the spectral data of the leaves. 990 cassava leaves and 330 roots were collected from 110 plants. The calibration and validation models were constructed by partial least square models with test set validation. High prediction performance models were found for predicting leaf dry matter (DM) and total cyanide (HCN) content using leaf spectra. The coefficient of determination of the validation set (R²ᵥₐₗ), root means square error of prediction (RMSEP), and the ratio performance deviation (RPDᵥₐₗ) of leaf DM were 0.891, 1.04%, and 3.03, respectively, and those of leaf HCN were 0.909, 60.5 ppm, and 3.32, respectively. The prediction model developed from root spectra showed an excellent prediction result for root DM (R²ᵥₐₗ = 0.979, RMSEP = 0.677%, RPDᵥₐₗ = 6.92), while good predicting ability models were obtained for HCN, starch, and TSS in root. Moreover, prediction models developed from leaf spectra could be used for qualitative screening of root quality. This study showed the potential of using NIRS as a rapid and reliable method to predict cassava quality at harvest or before processing.
Leaf diseases are a global threat to crop production and food preservation. Detecting these diseases is crucial for effective management. We introduce LeafDoc-Net, a robust, lightweight transfer-learning architecture for accurately detecting leaf diseases across multiple plant species, even with limited image data. Our approach concatenates two pre-trained image classification deep learning-based models, DenseNet121 and MobileNetV2. We enhance DenseNet121 with an attention-based transition mechanism and global average pooling layers, while MobileNetV2 benefits from adding an attention module and global average pooling layers. We deepen the architecture with extra-dense layers featuring swish activation and batch normalization layers, resulting in a more robust and accurate model for diagnosing leaf-related plant diseases. LeafDoc-Net is evaluated on two distinct datasets, focused on cassava and wheat leaf diseases, demonstrating superior performance compared to existing models in accuracy, precision, recall, and AUC metrics. To gain deeper insights into the model's performance, we utilize Grad-CAM++.
Introduction In the past years, it has been observed that the breeding of plants has become more challenging, as the visible difference in phenotypic data is much smaller than decades ago. With the ongoing climate change, it is necessary to breed crops that can cope with shifting climatic conditions. To select good breeding candidates for the future, phenotypic experiments can be conducted under climate-controlled conditions. Above-ground traits can be assessed with different optical sensors, but for the root growth, access to non-destructively measured traits is much more challenging. Even though MRI or CT imaging techniques have been established in the past years, they rely on an adequate infrastructure for the automatic handling of the pots as well as the controlled climate. Methods To address both challenges simultaneously, the non-destructive imaging of plant roots combined with a highly automated and standardized mid-throughput approach, we developed a workflow and an integrated scanning facility to study root growth. Our “ chamber #8 ” contains a climate chamber, a material flow control, an irrigation system, an X-ray system, a database for automatic data collection, and post-processing. The goals of this approach are to reduce the human interaction with the various components of the facility to a minimum on one hand, and to automate and standardize the complete process from plant care via measurements to root trait calculation on the other. The user receives standardized phenotypic traits and properties that were collected objectively. Results The proposed holistic approach allows us to study root growth of plants in a field-like substrate non-destructively over a defined period and to calculate phenotypic traits of root architecture. For different crops, genotypic differences can be observed in response to climatic conditions which have already been applied to a wide variety of root structures, such as potatoes, cassava, or corn. Discussion It enables breeders and scientists non-destructive access to root traits. Additionally, due to the non-destructive nature of X-ray computed tomography, the analysis of time series for root growing experiments is possible and enables the observation of kinetic traits. Furthermore, using this automation scheme for simultaneously controlled plant breeding and non-destructive testing reduces the involvement of human resources.
In agriculture, early detection and localization of plant diseases in time using deep learning techniques can help farmers contain the spread of plant diseases. In this work, we apply object detection models to identify and localize various categories of cassava plant leaf diseases. These include You Only Look Once (YOLO) as well as Generalized Efficient Layer Aggregation Network(GELAN) models. We applied YOLO v9-e, YOLO v9-c, as well as GELAN-e and GELAN-c models. The models were successfully trained using a custom cassava dataset. Several evaluation indicators that include precision, recall and mean average precision(mAP) were analysed result. The results have been compared with an earlier version of YOLO model and show an improvement in evaluation indicators reaching above 80% in the majority of diseases.
The quality of cassava in this study was indexed according to the starch content (SC) and dry matter content (DM). The NIR spectroscopy had the potential to measure changes in SC and DM of field-grown cassava. However, due to the non-homogenous nature of cassava root and its unpredictable shape, measuring it using NIR spectroscopy presents a challenge. Therefore, the model developed from different scanning method (horizontal and vertical lines) and surface (original surface, removed brownish outer layer, and fresh surface) were compared. To demonstrate that the model performance could improve by using a few significant wavelengths, its accuracy was compared using either the full wavelength or individual wavelengths selected by bootstrapping soft shrinkage analysis (BOSS) and competitive adaptive reweighted sampling (CARS) methods. Finally, the impact of physical properties (sample sections, peel thickness, varieties, and ages) on model error were determined. The prediction accuracy was found to depend on the scanning method. Since the model developed using horizontal spectra obviously obtained highest accuracy. The difference in surface had no effect on accuracy based on the horizontal spectra. Therefore, original surface scanning was chosen due to its non-destructive performance. The prediction model using the CARS-PLSR method was considered appropriate for predicting SC and BOSS-PLSR for predicting DM, providing R²cₐₗ, r²ᵥₐₗ, SEP, and ratio of prediction to standard deviation (RPD) of 0.69, 0.66, 3.18%, and 1.81, respectively; and 0.63, 0.54, 4.15%, and 1.50, respectively. In addition, the sample section, variety and peel thickness of the cassava root were not effected in the error of prediction.
Plant leaf infection recognition using supervised machine learning has emerged as a promising solution to address the pressing challenges in agriculture and plant pathology. Firstly, a wide-ranging dataset of real-time cassava images, comprising both infected and healthy plant leaves, is methodically amassed and accurately annotated. The real time dataset covered healthy cassava leaves, Cassava Bacterial Blight diseases, Cassava Brown Streak Disease diseases, Cassava Green Mottle diseases, Cassava Green Mottle diseases and Cassava Mosaic Disease. The data preprocessing phase involves an assessment of filtering, noise reduction, resizing, and extraction techniques to maintain uniformity and augment the dataset’s variability. Following this, relevant details are extracted from the preprocessed cassava frames, which are then utilized as inputs for the machine learning model. The approach of combining Leaf Intensity Vector, Principle Component Analysis, Gray Level Co-occurrence Matrix, and Support Vector Machine is adopted for the recognition of cassava leaf diseases. Using the polynomial and RBF kernel of SVM, KNN, Random Forests, and Decision Trees, the extracted suggested features are finally categorized. The proposed cassava leaf diseases classification system performs well, providing greater SVM RBF accuracy (98.4%) on models with disordered cassava leaves.
Abstract Mobile near infrared spectroscopy ( SCiO™ ) can offer quick, in-field phenotyping of cassava roots for pasting properties. However, validation is necessary to verify that reasonable expectations are established for the accuracy of a prediction model. In the context of an ongoing breeding effort, we investigated the use of an inexpensive, portable spectrometer that only records a portion (740–1070 nm) of the whole NIR spectrum to predict cassava pasting properties. Three machine-learning models, namely glmnet, lm, and gbm, implemented in the Caret package in R statistical program, were solely evaluated to approve one or two best models to move on with calibration and optimization. Based on calibration statistics (R 2 , RMSE and MAE), the best model was identified and further optimized. We found that model calibrations using glmnet provided the best model for breakdown viscosity, peak viscosity and pasting temperature. The glmnet model using the first derivative, peak viscosity had calibration and validation accuracy of R 2 = 0.56 and R 2 = 0.51 respectively while breakdown had calibration and validation accuracy of R 2 = 0.66 and R 2 = 0.66 respectively. We also found out that stacking of pre-treatments with Moving Average, Savitzky Golay, First Derivative, Second derivative and Standard Normal variate using glmnet model resulted in calibration and validation accuracy of R 2 = 0.65 and R 2 = 0.64 respectively for pasting temperature. The developed calibration model predicted the pasting properties of HQCF with sufficient accuracy. Therefore, SCiO™ can be reliably deployed in screening early-generation breeding materials for pasting properties.
Abstract Whitefly (Aleurotrachelus socialis Bondar) is a major pest causing significant economic losses in cassava production systems in North South America. It diminishes cassava's photosynthesis by colonizing leaves, directly feeding on phloem sap, or excreting substances that foster sooty mold growth, reducing the photosynthetic area. The most effective pest management approach is deploying natural resistance in the crop. Identifying germplasm with superior whitefly‐resistance (WFR) through phenotypic evaluation distinguishing it from whitefly‐susceptible responses requires an accurate, high‐throughput, quantitative phenotyping method. We developed Nymphstar, an image‐based phenotyping tool, as an ImageJ plugin, quantifying third‐ and fourth‐instar nymphs and their leaf area they occupy through red, green, and blue color space analysis. Using Nymphstar, we tested 19 cassava genotypes and classified their resistance to A. socialis. The plugin proved efficient, completing the analysis in 25.56 min on average for the entire dataset. In contrast, manual counting for the same set of images took 425.23 min on average averaging around 6.29 min/image. Nymphstar was ∼17 times faster showcasing its efficiency. To assess WFR in cassava germplasm, we conducted a full‐bench caging free‐choice assay. This approach enhanced whitefly colonization on each cassava genotype, providing an accurate representation of resistance/susceptible while reducing operator bias. Nymphstar is a rapid, precise tool for automated nymphs counting and leaf area quantification. It facilitates the large‐scale assessment of cassava resistance to whitefly, eliminating bias associated with field assessment and manual counting.
Background Cassava utilization for food and/or industrial products depends on inherent properties of root dry matter content (DMC) and the starch fraction of amylose content (AC). Accordingly, in the present study, near-infrared reflectance spectroscopy (NIRS) models were developed to aid breeding and selection of DMC and AC as critical industrial traits taking care of root sample preparation and cassava germplasm diversity available in Uganda. Results Upon undertaking calibrations and cross-validations, best models were adopted for validation. DMC in calibration samples ranged from 20 to 45 g 100g -1 , whereas, for amylose content, it ranged from 14 to 33 g 100g -1 . In the validation set, average DMC was 29.5 g 100g -1 , whereas, for amylose content, it was 24.64 g 100g -1 . For DMC, a modified partial least square regression model had regression coefficients (R 2 ) of 0.98 and 0.96, respectively, in the calibration and validation set. These were also associated with low bias (-0.018) and ratio of performance deviation that ranged from 4.7 to 5.0. In addition, standard error of prediction values ranged from 0.9 g 100g -1 to 1.06 g 100g -1 . For AC, the regression coefficient was 0.91 for the calibration set and 0.94 for the validation set. A bias equivalent to -0.03 and a ratio of performance deviation of 4.23 were observed. Conclusion These findings confirm the robustness of NIRS in the estimation of dry matter content and amylose content in cassava roots and thus justify its use in routine cassava breeding operations. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Leaf area (LA) and biomass are important agronomic indicators of the growth and health of plants. Conventional methods for measuring the LA can be challenging, time-consuming, costly, and laborious, especially for a large-scale study. A hybrid approach of cluster-based photography and modeling was, thus, developed herein to improve practicality. To this end, data on cassava palmate leaves were collected under various conditions to cover a spectrum of viable leaf shapes and sizes. A total of 1,899 leaves from 3 cassava genotypes and 5 cultivation conditions were first assigned into clusters by size, based on their length (L) and width (W). Next, 111 representative leaves from all clusters were photographed, and data from image-processing were subsequently used for model development. The model based on the product of L and W outperformed the rest (R2 = 0.9566, RMSE = 20.00). The hybrid model was successfully used to estimate the LA of greenhouse-grown cassava as validation. This represents a breakthrough in the search for efficient, practical phenotyping tools for LA estimation, especially for large-scale experiments or remote fields with limited machinery.
Freshness is an important parameter that is indexed in the quality assessment of commercial cassava tubers. Cassava tubers that are not fresh have reduced starch content. Therefore, in this study, we aimed to develop a new approach to detect cassava root deterioration levels using thermal imaging with machine learning (ML). An underlying assumption was that nonfresh cassava roots may have fermentation inside that causes a difference in the inner temperature of the tuber. This creates the opportunity for the deterioration level to be measured using thermal imaging. The features (pixel intensity and temperature) that were extracted from the region of interest (ROI) in the form of tuber thermal images were analyzed with ML. Linear discriminant analysis (LDA), k-nearest neighbor (kNN), support vector machine (SVM), decision tree, and ensemble classifiers were applied to establish the optimal classification modeling algorithms. The highest accuracy model was developed from thermal images of cassava roots captured in a darkroom under a control temperature of 25 °C in the measurement chamber. The LDA, SVM, and ensemble classifiers gave the best overall performance for the discrimination of cassava root deterioration levels, with an accuracy of 86.7%. Interestingly, under uncontrolled environmental conditions, the combination of thermal imaging plus ML gave results that were of lower accuracy but still acceptable. Thus, our work revealed that thermal imaging coupled with ML was a promising method for the nondestructive evaluation of cassava root deterioration levels.
Freshness is an important parameter that is indexed in the quality assessment of commercial cassava tubers. Cassava tubers that are not fresh have reduced starch content. Therefore, in this study, we aimed to develop a new approach to detect cassava root deterioration levels using thermal imaging with machine learning (ML). An underlying assumption was that nonfresh cassava roots may have fermentation inside that causes a difference in the inner temperature of the tuber. This creates the opportunity for the deterioration level to be measured using thermal imaging. The features (pixel intensity and temperature) that were extracted from the region of interest (ROI) in the form of tuber thermal images were analyzed with ML. Linear discriminant analysis (LDA), k-nearest neighbor (kNN), support vector machine (SVM), decision tree, and ensemble classifiers were applied to establish the optimal classification modeling algorithms. The highest accuracy model was developed from thermal images of cassava roots captured in a darkroom under a control temperature of 25 °C in the measurement chamber. The LDA, SVM, and ensemble classifiers gave the best overall performance for the discrimination of cassava root deterioration levels, with an accuracy of 86.7%. Interestingly, under uncontrolled environmental conditions, the combination of thermal imaging plus ML gave results that were of lower accuracy but still acceptable. Thus, our work revealed that thermal imaging coupled with ML was a promising method for the nondestructive evaluation of cassava root deterioration levels.
Cassava is an important crop for both the food and bio-energy industries, serving as a primary carbohydrate source and a staple food. Knowledge of the starch content (SC) is a key parameter index its quality. Non-destructive measurement of SC is needed to track the same tubers to study behaviours of SC accumulation in cassava tubers, which benefit for breeders in discovering a good variety and farmer to indicating harvesting period. This paper involves the prediction of cassava tuber starch content (SC) using multiple near infrared (NIR) spectrometers, aiming to measure SC in fresh cassava tuber. This study applies both portable NIR spectrometers at 570–1031nm and 860 – 1760 nm. The best results of the model which developed provided R²p and RPD were 0.69 and 1.80 for VIS-NIR region. Meanwhile, the Mid-NIR region provided poor performance with R²p and RPD of 0.46 and 1.36, respectively. Therefore, this study aims to assess whether it was better for predicting SC if the model was developed using combined selected significant wavelength from both Vis-NIR and Mid-NIR regions using machine learning. Genetics algorithm (GA) and variables important projection (VIP) were used for selecting significant wavelength. Selected wavelengths were combined and then generate calibration models via machine learning (ML). The results of the SC model developed using NIR spectra from selected wavelength with SV regression method had the highest performance, with R²c and RMSEc of 0.88 and 1.28%. Meanwhile, the R²p, RMSEp, and RPD of 0.74, 1.86% and 1.97, respectively. Then, the best calibration model was used to measure the unknown sample which corrected from a different harvest season. The external test set provided R²p, RMSEp were 0.88 and 1.38%. The results indicate that combination of NIR with different region can achieve.
In this work, we present a novel dataset composed of spectral data and images of cassava crops with and without diseases. Together with the description of the dataset, we describe the protocol to collect such data in a controlled environment and in an open field where pests are not controlled. Crop disease diagnosis has been done in the past through the analysis of plant images taken with a smartphone camera. However, in some cases, disease symptoms are not visible. Furthermore, for some cassava diseases, once symptoms have manifested on the aerial part of the plant, the root which is the edible part of the plant has been totally destroyed. The goal of collecting this multimodality of the crop disease is early intervention, following the hypothesis that diseased crops without visible symptoms can be detected using spectral information. We collected visible and near-infrared spectra captured from leaves infected with two common cassava diseases namely; Cassava Brown Streak Disease and Cassava Mosaic Disease, as well as from healthy plants. Together, we also captured leaf imagery data that corresponds to the spectral information. In our experiments, biochemical data is collected and taken as the ground truth. Finally, agricultural experts provided a disease score per plant leaf from 1 to 5, 1 representing healthy and 5 severely diseased. The process of disease monitoring and data collection took 19 and 15 consecutive weeks for screenhouse and open field, respectively, until disease symptoms were visibly seen by the human eye.
Artificial Intelligence (AI) has been evident in the agricultural sector recently. The objective of AI in agriculture is to control crop pests/diseases, reduce cost, and improve crop yield. In developing countries, the agriculture sector faces numerous challenges in the form of knowledge gap between farmers and technology, disease and pest infestation, lack of storage facilities, among others. In order to resolve some of these challenges, this paper presents crop pests/disease datasets sourced from local farms in Ghana. The dataset is presented in two folds; the raw images which consists of 24,881 images (6,549-Cashew, 7,508-Cassava, 5,389-Maize, and 5,435-Tomato) and augmented images which is further split into train and test sets. The latter consists of 102,976 images (25,811-Cashew, 26,330-Cassava, 23,657-Maize, and 27,178-Tomato), categorized into 22 classes. All images are de-identified, validated by expert plant virologists, and freely available for use by the research community.
Background The purpose of this study was to investigate the potential of hyperspectral imaging for the characterization of cooking quality parameters, dry matter content (DMC), water absorption (WAB), and texture in cassava genotypes contrasting for their cooking quality. Results Hyperspectral images were acquired on cooked and fresh intact longitudinal and transversal slices from 31 cassava genotypes harvested in March 2022 in Colombia. Different chemometric methods were tested for the quantification of DMC, WAB, and texture parameters. Data analysis was conducted through partial least squares regression, K nearest neighbors regression, support vector machine regression and CovSel multiple linear regression (CovSel_MLR). Efficient performances were obtained for DMC using CovSel_MLR with, coefficient of multiple determination Rp2=0.94 , root-mean-square error of prediction RMSEP = 0.96 g/100 g, and ratio of the standard deviation values RPD = 3.60. High heterogeneity was observed between contrasting genotypes. The predicted distribution of DMC within the root can be homogeneous or heterogeneous depending on the genotype. Weak predictions were obtained for WAB and texture parameters. Conclusions This study showed that hyperspectral imaging could be used as a high-throughput phenotyping tool for the visualization of DMC in contrasting cooking quality genotypes. Further improvement of protocols and larger datasets are required for WAB and texture quality traits. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Overlooked diseases in agriculture severely impact crop growth, which results in significant losses for farmers. Unfortunately, manual field visits for plant disease diagnosis (PDD) are costly and time consuming. Although various methods of PDD have been proposed, many challenges have yet to be investigated, such as early stage leaf disease diagnosis, class variations in diseases, cluttered backgrounds, and computational complexity of the diagnosis system. In this paper, we propose a Convolutional Neural Network (CNN)-based PDD framework (i.e., PDD-Net), which employs data augmentation techniques and incorporates multilevel and multiscale features to create a class and scale-invariant architecture. The Flatten-T Swish (FTS) activation function is utilized to prevent gradient vanishing and exploding problems, while the focal loss function is used to mitigate the impact of class imbalance during PDD-Net training. The PDD-Net method outperforms baseline models, achieving an average precision of 92.06%, average recall of 92.71%, average F1 score of 92.36%, and accuracy of 93.79% on the PlantVillage dataset. It also achieves an average precision of 86.41%, average recall of 85.77%, average F1 score of 86.02%, and accuracy of 86.98% on the cassava leaf disease dataset. These results demonstrate the efficiency and robustness of PDD-Net in plant disease diagnosis.
Background The non-invasive 3D-imaging and successive 3D-segmentation of plant root systems has gained interest within fundamental plant research and selectively breeding resilient crops. Currently the state of the art consists of computed tomography (CT) scans and reconstruction followed by an adequate 3D-segmentation process. Challenge Generating an exact 3D-segmentation of the roots becomes challenging due to inhomogeneous soil composition, as well as high scale variance in the root structures themselves. Approach (1) We address the challenge by combining deep convolutional neural networks (DCNNs) with a weakly supervised learning paradigm. Furthermore, (2) we apply a spatial pyramid pooling (SPP) layer to cope with the scale variance of roots. (3) We generate a fine-tuned training data set with a specialized sub-labeling technique. (4) Finally, to yield fast and high-quality segmentations, we propose a specialized iterative inference algorithm, which locally adapts the field of view (FoV) for the network. Experiments We compare our segmentation results against an analytical reference algorithm for root segmentation ( RootForce ) on a set of roots from Cassava plants and show qualitatively that an increased amount of root voxels and root branches can be segmented. Results Our findings show that with the proposed DCNN approach combined with the dynamic inference, much more, and especially fine, root structures can be detected than with a classical analytical reference method. Conclusion We show that the application of the proposed DCNN approach leads to better and more robust root segmentation, especially for very small and thin roots.
Background Gari and eba, forms of cassava semolina, are mainly consumed in Nigeria and other West African countries. This study aimed to define the critical quality traits of gari and eba, to measure their heritability, to define medium and high throughput instrumental methods for use by breeders, and to link the traits with consumer preferences. The definition of a food product's profiles, including its biophysical, sensory, and textural qualities, and the identification of the characteristics that determine its acceptability, are important if new genotypes are to be adopted successfully. Results Eighty cassava genotypes and varieties (three different sets) from the International Institute of Tropical Agriculture (IITA) research farm were used for the study. Participatory processing and consumer testing data on different types of gari and eba products were integrated to prioritize the traits preferred by processors and consumers. The color, sensory, and instrumental textural properties of these products were determined using standard analytical methods, and standard operating protocols (SOPs) developed by the RTBfoods project (Breeding Roots, Tubers, and Banana Products for End-user Preferences, https://rtbfoods.cirad.fr). There were significant (P Conclusions The color properties of gari and eba, together with instrumental measures of hardness and cohesiveness, are important quantitative discriminants of cassava genotypes. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
Abstract In Sub‐Saharan Africa, experts visually examine the plants and look for disease symptoms on the leaves to diagnose cassava diseases, a subjective method. Machine learning algorithms have been employed to quickly identify and classify crop diseases. In this study, we propose a model that integrates a transfer learning approach with a deep Gaussian convolutional neural network model. In this study, two pre‐trained transfer learning models were used, that is, MobileNet V2 and VGG16, together with three different kernels: a hybrid kernel (a product of a squared exponential kernel and a rational quadratic kernel), a squared exponential kernel, and a rational quadratic kernel. In experiments using MobileNet V2 and the three kernels, the hybrid kernel performed better, with an accuracy of 90.11%, compared to 86.03% and 85.14% for the squared exponential kernel and a rational quadratic kernel, respectively. Additionally, experiments using VGG16 and the three kernels showed that the hybrid kernel performed better, with an accuracy of 88.63%, compared to the squared exponential kernel's accuracy of 84.62% and the rational quadratic kernel's accuracy of 83.95%, respectively. All the experiments were done using a traditional computer with no access to GPU and this was the major limitation of the study.
Growing cassava in Africa requires resistance against the viruses causing cassava mosaic disease (CMD) and the viruses causing cassava brown streak disease (CBSD). A dominant CMD2 resistance gene from a West African cassava landrace provides strong resistance against the cassava mosaic viruses. However, resistance against cassava brown streak viruses is limited to cassava varieties that show tolerance to the disease. A recently identified cassava germplasm that cannot be infected with cassava brown streak viruses provides a new source of the resistance required to protect cassava from CBSD. We present a synopsis of the status of virus resistance in cassava and report on the research to combine resistance against CBSD and CMD. We improve the lengthy and erratic screening for CBSD resistance by proposing a virus infection and screening protocol for the viruses causing CBSD and CMD, which allows a rapid and precise assessment of cassava resistance under controlled conditions. Using this approach, we classified the virus responses of cassava lines from Africa and South America and identified truly virus-resistant clones that cannot be infected with any of the known viruses causing CBSD even under the most stringent virus infections. A modification of this protocol was used to test seedlings from cassava crosses for resistance against both diseases. A broad-spectrum resistance was identified in a workflow that lasted 9 months from seed germination to the identification of virus resistance. The workflow we propose dramatically reduces the evaluation and selection time required in a classical breeding workflow to reach the advanced field trial stage in only 9 months by conducting selections for virus resistance and plant multiplication in parallel. However, it does not bypass field evaluations; cassava resistance assessment prior to the field limits the evaluation to candidates with virus resistance defined as the absence of symptoms and the absence of the virus. The transfer of our virus screening workflow to cassava breeding programs enhances the efficiency by which resistance against viruses can be selected. It provides a precise definition of the plant's resistance response and can be used as a model system to tackle resistance in cassava against other diseases.
Cassava ( Manihot esculenta Crantz) starch consists of amylopectin and amylose, with its properties determined by the proportion of these two polymers. Waxy starches contain at least 95% amylopectin. In the food industry, waxy starches are advantageous, with pastes that are more stable towards retrogradation, while high-amylose starches are used as resistant starches. This study aimed to associate near-infrared spectrophotometry (NIRS) spectra with the waxy phenotype in cassava seeds and develop an accurate classification model for indirect selection of plants. A total of 1127 F 2 seeds were obtained from controlled crosses performed between 77 F 1 genotypes (wild-type, Wx _). Seeds were individually identified, and spectral data were obtained via NIRS using a benchtop NIRFlex N-500 and a portable SCiO device spectrometer. Four classification models were assessed for waxy cassava genotype identification: k-nearest neighbor algorithm (KNN), C5.0 decision tree (CDT), parallel random forest (parRF), and eXtreme Gradient Boosting (XGB). Spectral data were divided between a training set (80%) and a testing set (20%). The accuracy, based on NIRFlex N-500 spectral data, ranged from 0.86 (parRF) to 0.92 (XGB). The Kappa index displayed a similar trend as the accuracy, considering the lowest value for the parRF method (0.39) and the highest value for XGB (0.71). For the SCiO device, the accuracy (0.88-0.89) was similar among the four models evaluated. However, the Kappa index was lower than that of the NIRFlex N-500, and this index ranged from 0 (parRF) to 0.16 (KNN and CDT). Therefore, despite the high accuracy these last models are incapable of correctly classifying waxy and non-waxy clones based on the SCiO device spectra. A confusion matrix was performed to demonstrate the classification model results in the testing set. For both NIRS, the models were efficient in classifying non-waxy clones, with values ranging from 96-100%. However, the NIRS differed in the potential to predict waxy genotype class. For the NIRFlex N-500, the percentage ranged from 30% (parRF) to 70% (XGB). In general, the models tended to classify waxy genotypes as non-waxy, mainly SCiO. Therefore, the use of NIRS can perform early selection of cassava seeds with a waxy phenotype.
Cassava disease is one of the leading causes to the serious decline of cassava yield. Because it is difficult to identify the characteristics of cassava disease, if not professional cassava growers, it will be prone to misjudgment. In order to strengthen the judgment of cassava diseases, the identification characteristics of cassava diseases such as different color of cassava leaf disease spots, abnormal leaf shape and disease spot area were studied. In this paper, deep convolutional neural network was used to classify cassava leaf diseases, and image classification technology was used to recognize and classify cassava leaf diseases. A lightweight module Multi-scale fusion model (MSFM) based on attention mechanism was proposed to extract disease features of cassava leaves to enhance the classification of disease features. The resulting feature map contained key disease identification information. The study used 22,000 cassava disease leaf images as a data set, including four different cassava leaf disease categories and healthy cassava leaves. The experimental results show that the cassava leaf disease classification model based on multi-scale fusion Convolutional Neural Network (CNN) improves EfficientNet compared with the original model, with the average recognition rate increased by nearly 4% and the average recognition rate up to 88.1%. It provides theoretical support and practical tools for the recognition and early diagnosis of plant disease leaves.
Frequency is essential in signal transmission, especially in convolutional neural networks. It is vital to maintain the signal frequency in the neural network to maintain the performance of a convolutional neural network. Due to destructive signal transmission in convolutional neural network, signal frequency downconversion in channels results into incomplete spatial information. In communication theory, the number of Fourier series coefficients determines the integrity of the information transmitted in channels. Consequently, the number of Fourier series coefficients of the signals can be replenished to reduce the information transmission loss. To achieve this, the ArsenicNetPlus neural network was proposed for signal transmission modulation in detecting cassava diseases. First, multiattention was used to maintain the long-term dependency of the features of cassava diseases. Afterward, depthwise convolution was implemented to remove aliasing signals and downconvert before the sampling operation. Instance batch normalization algorithm was utilized to keep features in an appropriate form in the convolutional neural network channels. Finally, the ArsenicPlus block was implemented to generate pseudo high-frequency in the residual structure. The proposed method was tested on the Cassava Datasets and compared with the V2-ResNet-101, EfficientNet-B5, RepVGG-B3g4 and AlexNet. The results showed that the proposed method performed [Formula: see text] in terms of accuracy, 1.2440 in terms of loss, and [Formula: see text] in terms of the F1-score, outperforming the comparison algorithms.
The cassava starch market is promising in sub-Saharan Africa and increasing rapidly due to the numerous uses of starch in food industries. More accurate, high-throughput, and cost-effective phenotyping approaches could hasten the development of cassava varieties with high starch content to meet the growing market demand. This study investigated the effectiveness of a pocket-sized SCiO™ molecular sensor (SCiO) (740-1070 nm) to predict starch content in freshly ground cassava roots. A set of 344 unique genotypes from 11 field trials were evaluated. The predictive ability of individual trials was compared using partial least squares regression (PLSR). The 11 trials were aggregated to capture more variability, and the performance of the combined data was evaluated using two additional algorithms, random forest (RF) and support vector machine (SVM). The effect of pretreatment on model performance was examined. The predictive ability of SCiO was compared to that of two commercially available near-infrared (NIR) spectrometers, the portable ASD QualitySpec ® Trek (QST) (350-2500 nm) and the benchtop FOSS XDS Rapid Content™ Analyzer (BT) (400-2490 nm). The heritability of NIR spectra was investigated, and important spectral wavelengths were identified. Model performance varied across trials and was related to the amount of genetic diversity captured in the trial. Regardless of the chemometric approach, a satisfactory and consistent estimate of starch content was obtained across pretreatments with the SCiO (correlation between the predicted and the observed test set, ( R 2 P ): 0.84-0.90; ratio of performance deviation (RPD): 2.49-3.11, ratio of performance to interquartile distance (RPIQ): 3.24-4.08, concordance correlation coefficient (CCC): 0.91-0.94). While PLSR and SVM showed comparable prediction abilities, the RF model yielded the lowest performance. The heritability of the 331 NIRS spectra varied across trials and spectral regions but was highest ( H 2 > 0.5) between 871-1070 nm in most trials. Important wavelengths corresponding to absorption bands associated with starch and water were identified from 815 to 980 nm. Despite its limited spectral range, SCiO provided satisfactory prediction, as did BT, whereas QST showed less optimal calibration models. The SCiO spectrometer may be a cost-effective solution for phenotyping the starch content of fresh roots in resource-limited cassava breeding programs.
Storage roots of cassava plants crops are one of the main providers of starch in many South American, African, and Asian countries. Finding varieties with high yields is crucial for growing and breeding. This requires a better understanding of the dynamics of storage root formation, which is usually done by repeated manual evaluation of root types, diameters, and their distribution in excavated roots. We introduce a newly developed method that is capable to analyze the distribution of root diameters automatically, even if root systems display strong variations in root widths and clustering in high numbers. An application study was conducted with cassava roots imaged in a video acquisition box. The root diameter distribution was quantified automatically using an iterative ridge detection approach, which can cope with a wide span of root diameters and clustering. The approach was validated with virtual root models of known geometries and then tested with a time-series of excavated root systems. Based on the retrieved diameter classes, we show plausibly that the dynamics of root type formation can be monitored qualitatively and quantitatively. We conclude that this new method reliably determines important phenotypic traits from storage root crop images. The method is fast and robustly analyses complex root systems and thereby applicable in high-throughput phenotyping and future breeding.
In Sub-Saharan Africa, Professionals visually analyse the plants by looking for disease markers on the leaves to diagnose cassava infections, however, this method is extremely subjective. Automating the identification and classification of crop diseases may improve the accuracy of professional disease diagnosis and enable farmers in remote areas to monitor their crops without the assistance of experts. Algorithms for machine learning have been used in the early detection and classification of crop diseases. Motivated by the current developments in the field of Gaussian Processes, this study proposes to integrate the transfer learning approach with a deep Gaussian convolutional neural network model (DGCNN) for the detection and classification of cassava diseases. During this study, we used MobileNet V2 and VGG16 pre-trained transfer learning models and a hybrid kernel. Experiments with MobileNet V2 and a hybrid kernel revealed an accuracy of 90.11%. Also, experiments with VGG16 and a hybrid kernel revealed an accuracy of 88.63%. The major limitation of this study was computing resources since we used an ordinary computer in all our experiments. In our future work, we will experiment with the three kernel functions used in this study with kernel algorithms such as support vector machines and compare the results with those obtained during this study.
With the characteristic of high recognition rate and strong network robustness, convolutional neural network has now become the most mainstream method in the field of crop disease recognition. Aiming at the problems with insufficient numbers of labeled samples, complex backgrounds of sample images, and difficult extraction of useful feature information, a novel algorithm is proposed in this study based on attention mechanisms and convolutional neural networks for cassava leaf recognition. Specifically, a combined data augmentation strategy for datasets is used to prevent single distribution of image datasets, and then the PDRNet (plant disease recognition network) combining channel attention mechanism and spatial attention mechanism is proposed. The algorithm is designed as follows. Firstly, an attention module embedded in the network layer is deployed to establish remote dependence on each feature layer, strengthen the key feature information, and suppress the interference feature information, such as background noise. Secondly, a stochastic depth learning strategy is formulated to accelerate the training and inference of the network. And finally, a transfer learning method is adopted to load the pretrained weights into the model proposed in this study, with the recognition accuracy of the model enhanced by means of detailed parameter adjustments and dynamic changes in the learning rate. A large number of comparative experiments demonstrate that the proposed algorithm can deliver a recognition accuracy of 99.56% on the cassava disease image dataset, reaching the state-of-the-art level among CNN-based methods in terms of accuracy.
Background Methods to accurately quantify disease severity are fundamental to plant pathogen interaction studies. Commonly used methods include visual scoring of disease symptoms, tracking pathogen growth in planta over time, and various assays that detect plant defense responses. Several image-based methods for phenotyping of plant disease symptoms have also been developed. Each of these methods has different advantages and limitations which should be carefully considered when choosing an approach and interpreting the results. Results In this paper, we developed two image analysis methods and tested their ability to quantify different aspects of disease lesions in the cassava-Xanthomonas pathosystem. The first method uses ImageJ, an open-source platform widely used in the biological sciences. The second method is a few-shot support vector machine learning tool that uses a classifier file trained with five representative infected leaf images for lesion recognition. Cassava leaves were syringe infiltrated with wildtype Xanthomonas, a Xanthomonas mutant with decreased virulence, and mock treatments. Digital images of infected leaves were captured overtime using a Raspberry Pi camera. The image analysis methods were analyzed and compared for the ability to segment the lesion from the background and accurately capture and measure differences between the treatment types. Conclusions Both image analysis methods presented in this paper allow for accurate segmentation of disease lesions from the non-infected plant. Specifically, at 4-, 6-, and 9-days post inoculation (DPI), both methods provided quantitative differences in disease symptoms between different treatment types. Thus, either method could be applied to extract information about disease severity. Strengths and weaknesses of each approach are discussed.
Early disease identification in crops is critical for food security, especially in Sub-Saharan Africa. To identify cassava diseases, professionals visually score the plants by looking for disease indicators on the leaves which is notoriously subjective. Automating the detection and classification of crop diseases could help professionals diagnose diseases more accurately and allow farmers in remote locations to monitor their crops without the help of specialists. Machine learning algorithms have been used in the early detection and classification of crop diseases. However, traditional machine learning algorithms are not calibrated even though they have high accuracy. The ability to provide well-calibrated posterior distributions is one of the most appealing properties of Gaussian processes. Motivated by the current developments in the field of Gaussian Processes, this study proposed a deep Gaussian convolutional neural network model (DGCNN) for the detection and classification of cassava diseases using spectral data. The proposed model uses a hybrid kernel function that is the product of a rational quadratic kernel and a squared exponential kernel. Experimental results revealed that our proposed hybrid kernel function performed better in terms of accuracy of 90.10% when compared to both the squared exponential kernel with an accuracy of 88.01% and the rational quadratic kernel with an accuracy of 88.52%. In our future work, we propose to integrate the Optimised model proposed in this study with the transfer learning approach, a move that may help to improve the model performance.
Phenotypic analysis of cassava root crowns (CRCs) so far has been limited to visual inspection and very few measurements due to its laborious process in the field. Here, we developed a platform for acquiring 3D CRC models using close-range photogrammetry for phenotypic analysis. The state of the art is a low cost and easy to set up 3D acquisition requiring only a background sheet, a reference object and a camera, compatible with field experiments in remote areas. We tested different software with CRC samples, and Agisoft and Blender were the most suitable software for generating high-quality 3D models and data analysis, respectively. We optimized the workflow by testing different numbers of images for 3D reconstruction and found that a minimum of 25 images per CRC can provide high quality 3D models. Up to ten traits, including 3D crown volumes, 3D crown surface, root density, surface-to-volume ratio, root numbers, root angle, crown diameter, cylinder soil volume, CRC compactness and root length can be extracted providing novel parameters for studying cassava storage roots. We applied this platform to partial-inbred cassava populations and demonstrated that our platform provides reliable 3D CRC modelling for phenotypic analysis, analysis of genetic variances and supporting breeding selection.
BackgroundMethods to accurately quantify disease severity are fundamental to plant pathogen interaction studies. Commonly used methods include visual scoring of disease symptoms, tracking pathogen growth in planta over time, and various assays that detect plant defense responses. Several image-based methods for phenotyping of plant disease symptoms have also been developed. Each of these methods has different advantages and limitations which should be carefully considered when choosing an approach and interpreting the results. ResultsIn this paper, we developed two image analysis methods and tested their ability to quantify different aspects of disease lesions in the cassava-Xanthomonas pathosystem. The first method uses ImageJ, an open-source platform widely used in the biological sciences. The second method is a few-shot support vector machine learning tool that uses a classifier file trained with five representative infected leaf images for lesion recognition. Cassava leaves were syringe infiltrated with wildtype Xanthomonas, a Xanthomonas mutant with decreased virulence, and mock treatments. Digital images of infected leaves were captured overtime using a Raspberry Pi camera. The image analysis methods were analyzed and compared for the ability to segment the lesion from the background and accurately capture and measure differences between the treatment types. ConclusionsBoth image analysis methods presented in this paper allow for accurate segmentation of disease lesions from the non-infected plant. Specifically, at 4-, 6-, and 9-days post inoculation (DPI), both methods provided quantitative differences in disease symptoms between different treatment types. Thus, either method could be applied to extract information about disease severity. Strengths and weaknesses of each approach are discussed.
Abstract Cassava brown streak disease (CBSD) is an emerging viral disease that can greatly reduce cassava productivity, while causing only mild aerial symptoms that develop late in infection. Early detection of CBSD enables better crop management and intervention. Current techniques require laboratory equipment and are labour intensive and often inaccurate. We have developed a handheld active multispectral imaging (A-MSI) device combined with machine learning for early detection of CBSD in real-time. The principal benefits of A-MSI over passive MSI and conventional camera systems are improved spectral signal-to-noise ratio and temporal repeatability. Information fusion techniques further combine spectral and spatial information to reliably identify features that distinguish healthy cassava from plants with CBSD as early as 28 days post inoculation on a susceptible and a tolerant cultivar. Application of the device has the potential to increase farmers’ access to healthy planting materials and reduce losses due to CBSD in Africa. It can also be adapted for sensing other biotic and abiotic stresses in real-world situations where plants are exposed to multiple pest, pathogen and environmental stresses.
Cassava is a crucial food and nutrition security crop cultivated by small-scale farmers and it can survive in a brutal environment. It is a significant source of carbohydrates in African countries. Sometimes, Cassava crops can be infected by leaf diseases, affecting the overall production and reducing farmers’ income. The existing Cassava disease research encounters several challenges, such as poor detection rate, higher processing time, and poor accuracy. This research provides a comprehensive learning strategy for real-time Cassava leaf disease identification based on enhanced CNN models (ECNN). The existing Standard CNN model utilizes extensive data processing features, increasing the computational overhead. A depth-wise separable convolution layer is utilized to resolve CNN issues in the proposed ECNN model. This feature minimizes the feature count and computational overhead. The proposed ECNN model utilizes a distinct block processing feature to process the imbalanced images. To resolve the color segregation issue, the proposed ECNN model uses a Gamma correction feature. To decrease the variable selection process and increase the computational efficiency, the proposed ECNN model uses global average election polling with batch normalization. An experimental analysis is performed over an online Cassava image dataset containing 6256 images of Cassava leaves with five disease classes. The dataset classes are as follows: class 0: “Cassava Bacterial Blight (CBB)”; class 1: “Cassava Brown Streak Disease (CBSD)”; class 2: “Cassava Green Mottle (CGM)”; class 3: “Cassava Mosaic Disease (CMD)”; and class 4: “Healthy”. Various performance measuring parameters, i.e., precision, recall, measure, and accuracy, are calculated for existing Standard CNN and the proposed ECNN model. The proposed ECNN classifier significantly outperforms and achieves 99.3% accuracy for the balanced dataset. The test findings prove that applying a balanced database of images improves classification performance.
Image recognition plays a major role in everyday life applications like medical image analysis, gaming, surveillance and security, industrial automation, and more recently it has gained massive backing in the agricultural industry to identify plant diseases in crops. Plant diseases are a huge problem in agriculture and incorporating machine learning algorithms for their early detection will help better yields and save the farmers from loses. This paper entails the use of such machine learning algorithms to detect leaf diseases in the cassava plant. Cassava is one of the largest sources of carbohydrates for the continent of Africa. It is also very vulnerable to several plant diseases; this in turn threatens the food security of the continent. The present study is based on four of such diseases that affect the cassava yield namely, Cassava Bacterial Blight (CBB), Cassava Brown Streak Disease (CBSD), Cassava Mosaic Disease (CMD), and Cassava Green Mottle (CGM). In this research work, EfficientNet-B0 is proposed for the early detection of these diseases. The EfficientNet-B0 models outperform existing CNNs in terms of accuracy and efficiency while reducing parameter size and FLOPS by an order of magnitude. It is easier to detect disease at an early stage without the assistance of professionals, saving farmers both time and money. And our proposed model gave an accuracy of 92.6%.
Abstract In recent years, many countries prolonged their research in the field of agricultural farming for yielding more quantity food products with good quality. Agricultural farming is closely related to every country's economy in significant way. To improve the grain productivity in a notable way, plants must be protected from various viral, fungal, and bacterial diseases. The timely identification of plant disease helps to protect the plant at the early stage of the disease else it will spoil the entire plant. In this article, the plant disease identification problem is handled using our novel proposed network where it combines dilated convolution with residual dense block (DCRDB) along with multi‐level feature detection (MLFD) for selecting the appropriate feature and bidirectional long short‐term memory (Bi‐LSTM) classifier for leaf disease prediction. In DCRDB, the dilated convolution derives larger receptive field without loss of resolution to extract greater number of local features from the leaf images and residual dense block expands the size of feature map by densely connecting the layers of convolution. In addition, MLFD network integrates the high‐level features with the resulted features of DCRDB. After feature extraction, Bi‐LSTM is employed to recognize the leaf disease which supports additional training in both forward and backward direction. Through a set of collected cassava leaf images our research carries forward. The proposed model accuracy is verified with three different datasets such as cassava disease classification dataset, self‐made cassava disease dataset and plant village dataset. The performance of the proposed model is compared with some existing methods. Experimental results show that the highest F1‐score value of 95.49% attained by our proposed model for identifying cassava leaf disease.
Phenotyping to quantify the total carotenoids content (TCC) is sensitive, time-consuming, tedious, and costly. The development of high-throughput phenotyping tools is essential for screening hundreds of cassava genotypes in a short period of time in the biofortification program. This study aimed to (i) use digital images to extract information on the pulp color of cassava roots and estimate correlations with TCC, and (ii) select predictive models for TCC using colorimetric indices. Red, green and blue images were captured in root samples from 228 biofortified genotypes and the difference in color was analyzed using L*, a*, b*, hue and chroma indices from the International Commission on Illumination (CIELAB) color system and lightness. Colorimetric data were used for principal component analysis (PCA), correlation and for developing prediction models for TCC based on regression and machine learning. A high positive correlation between TCC and the variables b* (r = 0.90) and chroma (r = 0.89) was identified, while the other correlations were median and negative, and the L* parameter did not present a significant correlation with TCC. In general, the accuracy of most prediction models (with all variables and only the most important ones) was high (R2 ranging from 0.81 to 0.94). However, the artificial neural network prediction model presented the best predictive ability (R2 = 0.94), associated with the smallest error in the TCC estimates (root-mean-square error of 0.24). The structure of the studied population revealed five groups and high genetic variability based on PCA regarding colorimetric indices and TCC. Our results demonstrated that the use of data obtained from digital image analysis is an economical, fast, and effective alternative for the development of TCC phenotyping tools in cassava roots with high predictive ability.
Abstract Background Whitefly (Aleurotrachelus socialis Bondar) is an important pest causing high economic losses in cassava production systems in the north of South America. It reduces the plant’s photosynthesis by colonizing the cassava leaves and either directly feeding on their phloem sap or excreting substances that allow the growth of sooty mold and the consequent reduction of the photosynthetic area. The deployment of the crop’s natural resistance to this pest is the most effective approach to its management. Phenotypic evaluation to identify germplasm with superior whitefly-resistance (WFR) levels from that showing a whitefly susceptible (WFS) response will benefit from the availability of an accurate high-throughput, quantitative phenotyping method. Results We developed an accurate and efficient image-based phenotyping method (Nymphstar) to quantify the total number of third and fourth instar nymphs through red, green, and blue color space (RGB) image analysis as a plugin for ImageJ. Nymphstar estimates both the number of third and fourth instar nymphs and the percentage of leaf area they occupy. We tested 19 cassava genotypes and classified them after data analysis as resistant or susceptible to A. socialis attack. Benchmarking Nymphstar against manual nymph counts performed by a specialist revealed a highly significant correlation between direct nymph counts and those estimated using Nymphstar, which analyzed images and counted nymphs 150 times faster than by manual counts. Full-bench caging for a free-choice assay facilitated our assessment of WFR in the cassava germplasm and early replicated trials in the experimental population and enhanced the efficiency of whitefly (A. socialis) colonization on each cassava genotype to accurately depict the intensity of the resistance/susceptible response under a choice regime, simultaneously reducing human operator bias. Conclusions Nymphstar is a fast, accurate image analysis screening tool for the automated counting of nymphs and quantification of leaf area they occupy, allowing for the assessment of cassava resistance to whitefly on a large number of cassava plants in a glasshouse-based assay while avoiding the potential bias normally associated with field assessment and manual counting.
Abstract Background: Root phenotyping methods are of increasing importance as researchers seek to understand belowground productivity and breeders work to select for root traits. Effective non-destructive root phenotyping methods do not exist for bulked-root and tuber crops such as potato and cassava. Cassava is a tropical crop widely grown by subsistence farmers throughout the tropics and is the fourth most important staple food crop in the world, yet lags in research. It has an extensive growth period sometimes exceeding 12 months. Early maturity is a major goal for breeders, but the ability to select for it is hampered by the lack of non-destructive yield estimation methods. GPR is a tool with potential to aid in bulked root selection, but standard methods have yet to be developed. In this study, we demonstrate good practice in GPR estimation of root mass, which was used as a proxy for cassava root mass, and investigate the effect of soil water content on measurement. Results: Significant correlation between GPR data and daikon root mass was found for three of the five irrigation treatments. Correlation strength improved with increased soil water content and decreased variation of soil water content between plots. Pearson correlation coefficient varied from 0.53 – 0.79. Conclusions: GPR can be used to estimate bulked root mass. Wet soil can improve the predictive quality of GPR data, but water content needs to be homogeneous throughout the study site and period. Determining the optimal soil water content will require further research.
Cassava brown streak disease (CBSD) is an emerging viral disease that can greatly reduce cassava productivity, while causing only mild aerial symptoms that develop late in infection. Early detection of CBSD enables better crop management and intervention. Current techniques require laboratory equipment and are labour intensive and often inaccurate. We have developed a handheld active multispectral imaging (A-MSI) device combined with machine learning for early detection of CBSD in real-time. The principal benefits of A-MSI over passive MSI and conventional camera systems are improved spectral signal-to-noise ratio and temporal repeatability. Information fusion techniques further combine spectral and spatial information to reliably identify features that distinguish healthy cassava from plants with CBSD as early as 28 days post inoculation. Application of the device has the potential to increase farmers' access to healthy planting materials and reduce losses due to CBSD in Africa. It can also be adapted for sensing other biotic and abiotic stresses in real-world situations where plants are exposed to multiple pest, pathogen and environmental stresses.
There are various machine learning algorithms being implemented across the agricultural domain as well as other computer vision domains for the image classification problems as well as object detection problems. These algorithms work on feature extraction from the images. One of the most used algorithms is Convolutional Neural Network (CNN), which helps in feature extraction. Another method which is currently ruling the realm of machine learning is transfer learning, where the knowledge gained by machine while learning to solve one problem is applied for solving another problem. This paper demonstrates how various CNN architectures and transfer learning techniques can be applied for the disease detection in cassava plant.
ABSTRACT Over 800 million people across the tropics rely on cassava as a major source of calories. While the root dry matter content (RDMC) of this starchy root crop is important for both producers and consumers, characterization of RDMC by traditional methods is time-consuming and laborious for breeding programs. Alternate phenotyping methods have been proposed but lack the accuracy, cost, or speed ultimately needed for cassava breeding programs. For this reason, we investigated the use of a low-cost, handheld NIR spectrometer for field-based RDMC prediction in cassava. Oven-dried measurements of RDMC were paired with 21,044 scans of roots of 376 diverse clones from 10 field trials in Nigeria and grouped into training and test sets based on cross-validation schemes relevant to plant breeding programs. Mean partial least squares regression model performance ranged from R 2 p = 0.62 - 0.89 for within-trial predictions, which is within the range achieved with laboratory-grade spectrometers in previous studies. Relative to other factors, model performance was highly impacted by the inclusion of samples from the same environment in both the training and test sets. Random forest variable importance analysis of root spectra revealed increased importance in a region previously identified as predictive of water content in plants (~950 - 990 nm). With appropriate model calibration, the tested spectrometer will allow for field-based collection of spectral data with a smartphone for accurate RDMC prediction and potentially other quality traits, a step that could be easily integrated into existing harvesting workflows of cassava breeding programs. CORE IDEAS A low-cost, handheld near-infrared spectrometer was tested for phenotyping of cassava roots Plant breeding-relevant cross-validation schemes were used for predictions High prediction accuracies were achieved for cassava root dry matter content A spectral region predictive of plant water content was identified as important
For people in developing countries, cassava is a major source of calories and carbohydrates. However, Cassava Mosaic Disease (CMD) has become a major cause of concern among farmers in sub-Saharan Africa countries, which rely on cassava for both business and local consumption. The article proposes a novel deep residual convolution neural network (DRNN) for CMD detection in cassava leaf images. With the aid of distinct block processing, we can counterbalance the imbalanced image dataset of the cassava diseases and increase the number of images available for training and testing. Moreover, we adjust low contrast using Gamma correction and decorrelation stretching to enhance the color separation of an image with significant band-to-band correlation. Experimental results demonstrate that using a balanced dataset of images increases the accuracy of classification. The proposed DRNN model outperforms the plain convolutional neural network (PCNN) by a significant margin of 9.25% on the Cassava Disease Dataset from Kaggle.
The review aimed to identify the different high‐throughput phenotyping (HTP) techniques that used for quality evaluation in cassava and yam breeding programmes, and this has provided insights towards the development of metrics and their application in cassava and yam improvements. A systematic review of the published research articles involved the use of NIRS in analysing the quality traits of cassava and yam was carried out, and Scopus, Science Direct, Web of Sciences and Google Scholar were searched. The results of the review established that NIRS could be used in understanding the chemical constituents (carbohydrate, protein, vitamins, minerals, carotenoids, moisture, starch, etc.) for high‐throughput phenotyping. This study provides preliminary evidence of the application of NIRS as an efficient and affordable procedure for HTP. However, the feasibility of using mid‐infrared spectroscopy (MIRS) and hyperspectral imaging (HSI) in combination with the NIRS could be further studied for quality traits phenotyping.
Accurate assessments of the yield potential of cassava are needed to analyse yield gaps, define yield targets and set benchmarks for actual yields in Nigeria. This study evaluated the crop model LINTUL-Cassava under assumed potential growth and water-limited conditions in Nigeria. On-farm experiments were conducted at six locations across the three major cassava growing agro-ecologies of Western Africa (Tropical Rainforest – Ogoja and Ikom in Cross River state, Rainforest Transition Savanna – Ekpoma in Edo state and Guinea Savanna – Otukpo in Benue state) during two subsequent seasons (2016 – 2018). Treatments included fertilizer rates calculated to support the assumed potential yields of 90 t fresh storage root yield ha⁻¹ y⁻¹ (equivalent to 32 t DM ha⁻¹, produced in a growing season of 12 months). Light interception (LI) and leaf area index (LAI) were measured each month. The weights of leaves, stems and storage roots were measured at 4 and 8 months after planting and at harvest, and radiation use efficiency (RUE) calculated. The Edo experiment from 2016 was without drought stress and was used to parameterise LINTUL-Cassava and calibrate assimilate partitioning as function of temperature sums. The average fraction of light intercepted during the season was 80 %, with a light extinction coefficient of 0.67 and a RUE of 2.8 g DM MJ⁻¹ intercepted photosynthetically active radiation (IPAR). After calibration, the LINTUL-Cassava model described the crop growth and observed patterns of LAI well in the experiments in Cross River and Edo (2017). Simulated and observed storage root yield at 4 MAP (vegetative period), 8 MAP (mid-season) and at harvest were strongly correlated (R² of 0.92), with a RMSE of 4.93 t DM ha⁻¹. We ascertained that RUE of cassava was much higher than previously observed in Africa, with an average storage root yield of 39 ± 7 t DM ha⁻¹. Consequently, potential yields are greater and yield gaps larger than expected or previously reported. We conclude that the LINTUL-Cassava model can provide an adequate estimate of storage root yield across major cassava growing agroecological zones in Nigeria under rainfed conditions.
Nuru is a deep learning object detection model for diagnosing plant diseases and pests developed as a public good by PlantVillage (Penn State University), FAO, IITA, CIMMYT, and others. It provides a simple, inexpensive and robust means of conducting in-field diagnosis without requiring an internet connection. Diagnostic tools that do not require the internet are critical for rural settings, especially in Africa where internet penetration is very low. An investigation was conducted in East Africa to evaluate the effectiveness of Nuru as a diagnostic tool by comparing the ability of Nuru, cassava experts (researchers trained on cassava pests and diseases), agricultural extension officers and farmers to correctly identify symptoms of cassava mosaic disease (CMD), cassava brown streak disease (CBSD) and the damage caused by cassava green mites (CGM). The diagnosis capability of Nuru and that of the assessed individuals was determined by inspecting cassava plants and by using the cassava symptom recognition assessment tool (CaSRAT) to score images of cassava leaves, based on the symptoms present. Nuru could diagnose symptoms of cassava diseases at a higher accuracy (65% in 2020) than the agricultural extension agents (40-58%) and farmers (18-31%). Nuru's accuracy in diagnosing cassava disease and pest symptoms, in the field, was enhanced significantly by increasing the number of leaves assessed to six leaves per plant (74-88%). Two weeks of Nuru practical use provided a slight increase in the diagnostic skill of extension workers, suggesting that a longer duration of field experience with Nuru might result in significant improvements. Overall, these findings suggest that Nuru can be an effective tool for in-field diagnosis of cassava diseases and has the potential to be a quick and cost-effective means of disseminating knowledge from researchers to agricultural extension agents and farmers, particularly on the identification of disease symptoms and their management practices.
The review aimed to identify the different high-throughput phenotyping (HTP) techniques that used for quality evaluation in cassava and yam breeding programmes, and this has provided insights towards the development of metrics and their application in cassava and yam improvements. A systematic review of the published research articles involved the use of NIRS in analysing the quality traits of cassava and yam was carried out, and Scopus, Science Direct, Web of Sciences and Google Scholar were searched. The results of the review established that NIRS could be used in understanding the chemical constituents (carbohydrate, protein, vitamins, minerals, carotenoids, moisture, starch, etc.) for high-throughput phenotyping. This study provides preliminary evidence of the application of NIRS as an efficient and affordable procedure for HTP. However, the feasibility of using mid-infrared spectroscopy (MIRS) and hyperspectral imaging (HSI) in combination with the NIRS could be further studied for quality traits phenotyping.
This paper involves the prediction of cassava tuber starch content (SC) using near-infrared (NIR) spectroscopy, aiming to follow the change of SC in individual tubers utilised for a breeding programme. This study applies a portable NIR spectrometer at wavelengths of 570–1031 nm in the evaluation of SC in fresh cassava tubers. The prediction models are established using partial least squares (PLS) regression with NIR spectra obtained in the interactance mode. The effective model was developed from the wavelength region of 600–1000 nm with spectral pre-processing of the second derivative, giving the coefficient of determination of prediction set (r²) and root mean square error of prediction (RMSEP) of 0.62 and 2.21%, respectively. The effect of tuber section (including head, middle and tail) on the performance of the SC model was investigated. The individual head, middle and tail models were acceptable for screening. However, the performances of the combined model (which is the model developed a mix of all individual section samples) and the individual section model were not significantly different. Therefore, the combined model was suitable in real application because of the ease of in-field scanning. The result demonstrates that the SCs of cassava tubers can be measured by a NIR spectroscopy method. Furthermore, it can be used as an alternative tool which is appropriate for breeders to use to follow the behaviour of SC during breeding.
Societal Impact Statement Micronutrient deficiency or “hidden hunger” is estimated to affect two billion people worldwide and increasing the micronutrient concentration of food could play an important role in tackling this global challenge. Using a combination of imaging techniques and atomic absorption spectroscopy, we describe a link between root phenotype and micronutrient concentration in cassava, which could enable new phenotypic selection strategies for breeding. This approach could be used with existing breeding infrastructure to enhance the micronutrient concentration of cassava and hence, benefit the health of people, particularly in low‐income countries where cassava is consumed as a staple crop. Summary Cassava storage roots are a staple food in low‐income countries of South‐East Asia and sub‐Saharan Africa, where growth stunting is prevalent as a consequence of micronutrient deficiencies. We aim to link phenotypes of field‐grown cassava roots to micronutrient concentration in the edible storage roots as a simple way to improve phenotypic selection for nutritional value in cassava. We used existing and newly developed imaging techniques to quantify root phenotypes of the cassava root architecture over time and used flame atomic absorption spectroscopy to measure micronutrient concentration in storage roots. Both together allow the association of root phenotypes with micronutrient concentration in mature cassava roots. We show that early and late bulking genotypes in cassava exhibit distinct foraging behaviors that are associated with micronutrient concentration in the edible storage root. Our observations suggest that late bulking cassava is a key to provide sufficient micronutrients in the edible storage root. The association between root phenotype and micronutrient concentration with imaging techniques allows phenotypic selection for enhanced micronutrient concentration. Therefore, implementing image‐based phenotyping into cassava breeding programs in sub‐Saharan Africa and South‐East Asia could be an essential element to resolve micronutrient deficiencies that puts individuals at a higher risk of growth stunting.
Short-wavelength near infrared spectra in the interactance mode were collected from intact cassava roots and cassava flesh, using two portable spectrometers for the spectral regions of 720-1050 and 850-1150 nm, respectively. All starch prediction models were developed using the partial least squares regression. Good prediction performance was obtained from the cassava flesh (cross-section cut root) measurement with a correlation of prediction ( r p ) of 0.917 and standard error of prediction (SEP) of 1.73%, for both spectrometers. For the intact root, the prediction models were satisfactorily accurate with r p values of 0.687 and 0.772 and SEP of 3.151 and 2.803%, respectively. Moreover, the performance measurement of all optimum models was also evaluated according to ISO 12099:2017(E). The results showed that the predicted values were not significantly different from the actual values obtained from the standard method at 95% confidence intervals. These results showed the feasibility of using portable spectrometers to predict the starch content of fresh cassava roots.
Abstract Background Rapid non-destructive measurements to predict cassava root yield over the full growing season through large numbers of germplasm and multiple environments is a huge challenge in Cassava breeding programs. As opposed to waiting until the harvest season, multispectral imagery using unmanned aerial vehicles (UAV) are capable of measuring the canopy metrics and vegetation indices (VIs) traits at different time points of the growth cycle. This resourceful time series aerial image processing with appropriate analytical framework is very important for the automatic extraction of phenotypic features from the image data. Many studies have demonstrated the usefulness of advanced remote sensing technologies coupled with machine learning (ML) approaches for accurate prediction of valuable crop traits. Until now, Cassava has received little to no attention in aerial image-based phenotyping and ML model testing. Results To accelerate image processing, an automated image-analysis framework called CIAT Pheno-i was developed to extract plot level vegetation indices/canopy metrics. Multiple linear regression models were constructed at different key growth stages of cassava, using ground-truth data and vegetation indices obtained from a multispectral sensor. Henceforth, the spectral indices/features were combined to develop models and predict cassava root yield using different Machine learning techniques. Our results showed that (1) Developed CIAT pheno-i image analysis framework was found to be easier and more rapid than manual methods. (2) The correlation analysis of four phenological stages of cassava revealed that elongation (EL) and late bulking (LBK) were the most useful stages to estimate above-ground biomass (AGB), below-ground biomass (BGB) and canopy height (CH). (3) The multi-temporal analysis revealed that cumulative image feature information of EL + early bulky (EBK) stages showed a higher significant correlation ( r = 0.77) for Green Normalized Difference Vegetation indices (GNDVI) with BGB than individual time points. Canopy height measured on the ground correlated well with UAV (CHuav)-based measurements ( r = 0.92) at late bulking (LBK) stage. Among different image features, normalized difference red edge index (NDRE) data were found to be consistently highly correlated ( r = 0.65 to 0.84) with AGB at LBK stage. (4) Among the four ML algorithms used in this study, k-Nearest Neighbours (kNN), Random Forest (RF) and Support Vector Machine (SVM) showed the best performance for root yield prediction with the highest accuracy of R 2 = 0.67, 0.66 and 0.64, respectively. Conclusion UAV platforms, time series image acquisition, automated image analytical framework (CIAT Pheno-i), and key vegetation indices (VIs) to estimate phenotyping traits and root yield described in this work have great potential for use as a selection tool in the modern cassava breeding programs around the world to accelerate germplasm and varietal selection. The image analysis software (CIAT Pheno-i) developed from this study can be widely applicable to any other crop to extract phenotypic information rapidly.
This study compared the relationship of different phenotyping methods including iCheckTM CAROTENE (iCheck), Chromameter, colour chart and visible/near-infrared spectroscopy (Vis/NIRS) used in quantifying total carotenoids content (TCC) in fresh cassava roots. Using a total of 194 cassava clones harvested from the International Institute of Tropical Agriculture (IITA), Ibadan, we compared the repeatability precision, accuracy of measurement and correlations of these phenotyping methods. From the results, Vis/NIRS-analyzed TCC had high and positive correlations with Chromameter and Color chart (r = 0.91 and 0.71, respectively). On the other hand, the result revealed somewhat moderate correlation (r = 0.67) between Vis/NIRS and iCheck measurements. Vis/NIRS, iCheck and chromameter methods gave high and nearly equal heritability estimates (0.95, 0.98 and 0.98, respectively) illustrating high repeatability precision of these methods; an indication that they can be used for germplasm selection in the early stages of breeding. Conversely, with Bland-Altman plot at 95% confidence level, the accuracy of iCheck was not comparable with that of Vis/ NIRS. The information derived from this analysis directly contributes towards the genetic improvement of root quality traits in cassava and facilitates the sharing of data across cassava breeding consortium.
Abstract Background: Rapid non-destructive measurements to predict cassava root yield over the full growing season through large numbers of germplasm and multiple environments is a huge challenge in Cassava breeding programs. As opposed to waiting until the harvest season, multispectral imagery using unmanned aerial vehicles (UAV) are capable of measuring the canopy metrics and vegetation indices (VIs) traits at different time points of the growth cycle. This resourceful time series aerial image processing with appropriate analytical framework is very important for the automatic extraction of phenotypic features from the image data. Many studies have demonstrated the usefulness of advanced remote sensing technologies coupled with machine learning (ML) approaches for accurate prediction of valuable crop traits. Until now, Cassava has received little to no attention in aerial image-based phenotyping and ML model testing. Results : To accelerate image processing, an automated image-analysis framework called CIAT Pheno-i was developed to extract plot level vegetation indices/canopy metrics. Multiple linear regression models were constructed at different key growth stages of cassava, using ground-truth data and vegetation indices obtained from a multispectral sensor. Henceforth, the spectral indices/features were combined to develop models and predict cassava root yield using different Machine learning techniques. Our results showed that (1) Developed CIAT pheno-i image analysis framework was found to be easier and more rapid than manual methods. (2) The correlation analysis of four phenological stages of cassava revealed that elongation (EL) and late bulking (LBK) were the most useful stages to estimate above-ground biomass (AGB), below-ground biomass (BGB) and canopy height (CH). (3) The multi-temporal analysis revealed that cumulative image feature information of EL+early bulky (EBK) stages showed a higher significant correlation ( r = 0.77) for Green Normalized Difference Vegetation indices (GNDVI) with BGB than individual time points. Canopy height measured on the ground correlated well with UAV (CHuav)-based measurements ( r = 0.92) at late bulking (LBK) stage. Among different image features, normalized difference red edge index (NDRE) data were found to be consistently highly correlated ( r = 0.65 to 0.84) with AGB at LBK stage. (4) Among the four ML algorithms used in this study, k-Nearest Neighbours (kNN), Random Forest (RF) and Support Vector Machine (SVM) showed the best performance for root yield prediction with the highest accuracy of R 2 = 0.67, 0.66 and 0.64, respectively. Conclusion : UAV platforms, time series image acquisition, automated image analytical framework (CIAT Pheno-i), and key vegetation indices (VIs) to estimate phenotyping traits and root yield described in this work have great potential for use as a selection tool in the modern cassava breeding programs around the world to accelerate germplasm and varietal selection. The image analysis software (CIAT Pheno-i) developed from this study can be widely applicable to any other crop to extract phenotypic information rapidly.
Cassava is cultivated due to its drought tolerance and high carbohydrate-containing storage roots. The lack of uniformity and irregular shape of storage roots poses constraints on harvesting and post-harvest processing. Here, we phenotyped the Genetic gain and offspring (C1) populations from the International Institute of Tropical Agriculture (IITA) breeding program using image analysis of storage root photographs taken in the field. In the genome-wide association analysis (GWAS), we detected for most shape and size-related traits, QTL on chromosomes 1 and 12. In a previous study, we found the QTL on chromosome 12 to be associated with cassava mosaic disease (CMD) resistance. Because the root uniformity is important for breeding, we calculated the standard deviation (SD) of individual root measurements per clone. With SD measurements we identified new significant QTL for Perimeter, Feret and Aspect Ratio on chromosomes 6, 9 and 16. Predictive accuracies of root size and shape image-extracted traits were mostly higher than yield trait prediction accuracies. This study aimed to evaluate the feasibility of the image phenotyping protocol and assess GWAS and genomic prediction for size and shape image-extracted traits. The methodology described and the results are promising and open up the opportunity to apply high-throughput methods in cassava.
Abstract Background: Rapid non-destructive measurements to predict cassava root yield over the full growing season through large numbers of germplasm and multiple environments is a huge challenge in Cassava breeding programs. As opposed to waiting until the harvest season, multispectral imagery using unmanned aerial vehicles (UAV) are capable of measuring the canopy metrics and vegetation indices (VIs) traits at different time points of the growth cycle. This resourceful time series aerial image processing with appropriate analytical framework is very important for the automatic extraction of phenotypic features from the image data. Many studies have demonstrated the usefulness of advanced remote sensing technologies coupled with machine learning (ML) approaches for accurate prediction of valuable crop traits. Until now, Cassava has received little to no attention in aerial image-based phenotyping and ML model testing. Results: To accelerate image processing, an automated image-analysis framework called CIAT Pheno-i was developed to extract plot level vegetation indices/canopy metrics. Multiple linear regression models were constructed at different key growth stages of cassava, using ground-truth data and vegetation indices obtained from a multispectral sensor. Henceforth, the spectral indices/features were combined to develop models and predict cassava root yield using different Machine learning techniques. Our results showed that (1) Developed CIAT pheno-i image analysis framework was found to be easier and more rapid than manual methods. (2) The correlation analysis of four phenological stages of cassava revealed that elongation (EL) and late bulking (LBK) were the most useful stages to estimate above-ground biomass (AGB), below-ground biomass (BGB) and canopy height (CH). (3) The multi-temporal analysis revealed that cumulative image feature information of EL+early bulky (EBK) stages showed a higher significant correlation (r = 0.77) for Green Normalized Difference Vegetation indices (GNDVI) with BGB than individual time points. Canopy height measured on the ground correlated well with UAV (CHuav)-based measurements (r = 0.92) at late bulking (LBK) stage. Among different image features, normalized difference red edge index (NDRE) data were found to be consistently highly correlated (r = 0.65 to 0.84) with AGB at LBK stage. (4) Among the four ML algorithms used in this study, k-Nearest Neighbours (kNN), Random Forest (RF) and Support Vector Machine (SVM) showed the best performance for root yield prediction with the highest accuracy of R2 = 0.67, 0.66 and 0.64, respectively. Conclusion: UAV platforms, time series image acquisition, automated image analytical framework (CIAT Pheno-i), and key vegetation indices (VIs) to estimate phenotyping traits and root yield described in this work have great potential for use as a selection tool in the modern cassava breeding programs around the world to accelerate germplasm and varietal selection. The image analysis software (CIAT Pheno-i) developed from this study can be widely applicable to any other crop to extract phenotypic information rapidly.
High‐throughput phenotyping tools that allow the early and accurate evaluation of important agronomic traits have gained space in current breeding programs. The aim of this study was to evaluate the potential of Fourier‐transform near‐infrared spectroscopy (FT‐NIRS) to identify cassava (Manihot esculenta Crantz) clones with waxy starch (i.e., amylose‐free) by screening leaves rather than roots, and to validate prediction models for classifying these phenotypes. We analyzed the spectra of 162 waxy and 180 nonwaxy genotypes from five different growing environments. The mean FT‐NIRS spectra and principal component analysis (PCA) were used to investigate the potential for grouping the data. For classification, five supervised pattern recognition techniques were tested: Bayesian generalized linear model (BGLM), high‐dimensional discriminant analysis (HDDA), partial least squares‐discriminant analysis (PLS‐DA), parallel random forest (PRANDF), and support vector machines with linear kernel (SVM). The mean spectra and the PCA did not allow discrimination of the genotypes based on starch classification. The SVM and BGLM showed the highest classification accuracy in cross‐validation (.86–.87), with higher concordance rates (.88–.83), sensitivity (.87–.85) and specificity (.88). The BGLM and SVM models also obtained better indices in the external validation, with high accuracy (.85) and correct classification of 93% of the waxy genotypes. Thus, performing early selection of root characteristics based on the indirect selection of variables extracted from leaf spectra is a good potential strategy for more efficient breeding of the waxy phenotype.
Abstract has not been obtained from indexed metadata or an accessible article page.
Cassava frog skin disease (CFSD) causes significant yield losses in cassava (Manihot esculenta Crantz). One issue with control is late diagnosis. The objective of this work was to test near-infrared spectrometry (NIRS) and compare calibration and classification models for early detection of CFSD. NIRS analysis was performed on 238 cassava accessions (120 healthy and 118 with CFSD). Six classification models were used: a support vector machine with a linear kernel (SVM), a Bayesian generalized linear model (BGLM), a parallel random forest (PRANDF), an extreme learning machine (ELM), a high dimensional discriminant analysis (HDDA), and a partial least squares (PLS) model. Their predictive abilities were evaluated based on their accuracy and agreement (based on Cohen’s Kappa coefficient). The models demonstrated high efficiency at distinguishing diseased and healthy accessions, with an overall accuracy >80%. The Cohen’s Kappa coefficient values (0.83–0.98) for four of the models (the SVM, BGLM, PRANDF and PLS) indicated that there was almost perfect agreement between their classification results. The SVM and BGLM models exhibited high accuracy (99.07 and 98.92%, respectively) and reproducibility (both 0.98) in classifying the accessions according to their CFSD status. NIRS is a viable alternative for the detection of CFSD with the advantages of early and accurate detection, high speed and low cost compared to traditional diagnostic methods.
Random forests (RF) was used to correlate spectral responses to known wet chemistry carotenoid concentrations including total carotenoid content (TCC), all-trans β-carotene (ATBC), violaxanthin (VIO), lutein (LUT), 15-cis beta-carotene (15CBC), 13-cis beta-carotene (13CBC), alpha-carotene (AC), 9-cis beta-carotene (9CBC), and phytoene (PHY) from laboratory analysis of 173 cassava root samples in Columbia. The cross-validated correlations between the actual and estimated carotenoid values using RF ranged from 0.62 in PHY to 0.97 in ATBC. The developed models were used to evaluate the carotenoids of 594 cassava clones with spectral information collected across three locations in a national breeding program (NRCRI, Umudike), Nigeria. Both populations contained cassava clones characterized as white and yellow. The NRCRI evaluated phenotypes were used to assess the genetic correlations, conduct genome-wide association studies (GWAS), and genomic predictions. Estimates of genetic correlation showed various levels of the relationship among the carotenoids. The associations between TCC and the individual carotenoids were all significant (P 0.75, except in LUT and PHY where r < 0.3). The GWAS revealed significant genomic regions on chromosomes 1, 2, 4, 13, 14, and 15 associated with variation in at least one of the carotenoids. One of the identified candidate genes, phytoene synthase (PSY) has been widely reported for variation in TCC in cassava. On average, genomic prediction accuracies from the single-trait genomic best linear unbiased prediction (GBLUP) and RF as well as from a multiple-trait GBLUP model ranged from ∼0.2 in LUT and PHY to 0.52 in TCC. The multiple-trait GBLUP model gave slightly higher accuracies than the single trait GBLUP and RF models. This study is one of the initial attempts in understanding the genetic basis of individual carotenoids and demonstrates the usefulness of NIRS in cassava improvement.
Cassava roots are complex structures comprising several distinct types of root. The number and size of the storage roots are two potential phenotypic traits reflecting crop yield and quality. Counting and measuring the size of cassava storage roots are usually done manually, or semi-automatically by first segmenting cassava root images. However, occlusion of both storage and fibrous roots makes the process both time-consuming and error-prone. While Convolutional Neural Nets have shown performance above the state-of-the-art in many image processing and analysis tasks, there are currently a limited number of Convolutional Neural Net-based methods for counting plant features. This is due to the limited availability of data, annotated by expert plant biologists, which represents all possible measurement outcomes. Existing works in this area either learn a direct image-to-count regressor model by regressing to a count value, or perform a count after segmenting the image. We, however, address the problem using a direct image-to-count prediction model. This is made possible by generating synthetic images, using a conditional Generative Adversarial Network (GAN), to provide training data for missing classes. We automatically form cassava storage root masks for any missing classes using existing ground-truth masks, and input them as a condition to our GAN model to generate synthetic root images. We combine the resulting synthetic images with real images to learn a direct image-to-count prediction model capable of counting the number of storage roots in real cassava images taken from a low cost aeroponic growth system. These models are used to develop a system that counts cassava storage roots in real images. Our system first predicts age group ('young' and 'old' roots; pertinent to our image capture regime) in a given image, and then, based on this prediction, selects an appropriate model to predict the number of storage roots. We achieve 91% accuracy on predicting ages of storage roots, and 86% and 71% overall percentage agreement on counting 'old' and 'young' storage roots respectively. Thus we are able to demonstrate that synthetically generated cassava root images can be used to supplement missing root classes, turning the counting problem into a direct image-to-count prediction task.
Abstract Background Root and tuber crops are becoming more important for their high source of carbohydrates, next to cereals. Despite their commercial impact, there are significant knowledge gaps about the environmental and inherent regulation of storage root (SR) differentiation, due in part to the innate problems of studying storage roots and the lack of a suitable model system for monitoring storage root growth. The research presented here aimed to develop a reliable, low-cost effective system that enables the study of the factors influencing cassava storage root initiation and development. Results We explored simple, low-cost systems for the study of storage root biology. An aeroponics system described here is ideal for real-time monitoring of storage root development (SRD), and this was further validated using hormone studies. Our aeroponics-based auxin studies revealed that storage root initiation and development are adaptive responses, which are significantly enhanced by the exogenous auxin supply. Field and histological experiments were also conducted to confirm the auxin effect found in the aeroponics system. We also developed a simple digital imaging platform to quantify storage root growth and development traits. Correlation analysis confirmed that image-based estimation can be a surrogate for manual root phenotyping for several key traits. Conclusions The aeroponic system developed from this study is an effective tool for examining the root architecture of cassava during early SRD. The aeroponic system also provided novel insights into storage root formation by activating the auxin-dependent proliferation of secondary xylem parenchyma cells to induce the initial root thickening and bulking. The developed system can be of direct benefit to molecular biologists, breeders, and physiologists, allowing them to screen germplasm for root traits that correlate with improved economic traits.
Cassava is widely cultivated due to its high drought tolerance and high carbohydrate-containing storage roots. The lack of uniformity and irregular shape of storage roots within and between genotypes poses significant constraints on harvesting and post-harvest processing. Routine assessment of storage root size and shape in breeding plots relies on visual scores. Here, we phenotyped the Genetic gain and offspring (C1) populations from the International Institute of Tropical Agriculture (IITA) breeding program for root shape and size-related traits using image analysis of storage root photographs taken in the field. In our study, using univariate genome-wide association analysis, we detected for most shape and size related traits, significant QTL regions located on chromosomes 1 and 12. The QTL region on chromosome 12 has been previously associated, using IITA breeding populations, to cassava mosaic disease (CMD) resistance. Because the uniformity in size and shape of cassava roots is an important breeding goal, we calculated the standard deviation of individual root measurements per clone. The use of standard deviation measurements allowed the identification of new significant QTL for Perimeter, Feret and Aspect Ratio on chromosomes 6, 9 and 16. Using genomic prediction cross validation, the accuracies of root size and shape-related traits were lower than those previously reported for dry matter content (DM) and cassava mosaic virus resistance (CMD). Predictive accuracies of the mean values of root size and shape image-extracted traits were mostly higher than yield trait prediction accuracies in the C1 population. This study aimed to evaluate the feasibility of the image phenotyping protocol and to assess the use of genome-wide analyses for size and shape image-extracted traits. The methodology described here and the results obtained in this study are promising and open up the opportunity to apply high-throughput methods in cassava.
The development of cassava (Manihot esculenta Crantz) varieties with greater tolerance of water deficit depends on optimized phenotyping tools. The objective of this work was to develop early prediction models of final root yield (12 months after planting - MAP) using physiological and agronomic data obtained at 4 MAP under two water regimes. Nine genotypes of cassava were evaluated under two treatments (irrigated and with water deficit), using a complete randomized block design, in a factorial scheme of 2 harvest periods (at 4 and 12 MAP) × 9 genotypes, with four replications. Both treatment groups were irrigated until 3 MAP. After this period, irrigation was interrupted for the water deficit treatment group. Fourteen physiological and agronomic traits were evaluated in all harvest periods. Four prediction models were evaluated: linear regression with stepwise selection (LRSS), linear regression with backward selection (LRBS), Bayesian ridge regression (BRR), and partial least squares (PLS). Most of the models presented a high predictive ability for final root yield (R2 ranging from 0.83 to 0.91). However, in all prediction scenarios, the PLS model presented a high R2 (0.84 to 0.91) associated with the lowest root-mean-square error (RMSE) (0.82 to 1.60). Differences in the predictive ability of the models may have occurred due to the relative importance of the early traits. In the case of PLS, the most important traits for the model were stomatal conductance, root yield at 4 MAP, leaf area index and number of roots. Regardless of the water condition, the physiological and agronomic data collected at an early stage could successfully be used to predict the final root yield with great efficiency. This strategy can reduce the cost of phenotyping, increasing the capacity for analysis and optimization of genetic gains for tolerance to drought in cassava.
Although breeders have made significant progress in the genetic improvement of cassava (Manihot esculenta Crantz) for agronomic traits, lack of information on heritability and limited testing of high-throughput phenotyping methods are major limitations to improving root quality traits, such as softness after cooking, which rank high among Ugandan consumers. The objectives of this study were to determine heritability for softness of cooked cassava roots, and quantify the relationship between penetrometer and consumer testing methods for phenotyping softness of cassava roots. Softness defined as the maximum force (N) needed to penetrate cooked root samples using a penetrometer, was evaluated at four cooking time intervals: 15, 30, 45, and 60 min on 268 cassava genotypes. Estimates of broad-sense heritability (repeatability) ranged from 0.17 to 0.37, with the highest value observed at 45 min of cooking time interval. In the second study involving 135 cassava consumers from Kibaale district in Uganda, penetrometer measurements of cooked roots from six cassava varieties were found to be in strong agreement (r2 = 0.91; P-value = 0.003) with ordinal scores of root softness from consumer testing. These results suggest that: (a) softness of cooked cassava roots is a trait amenable for evaluation and selection; and (b) a penetrometer can readily be used for assessment of cooked root softness. These findings form the basis for operationalising the routine assessment of root softness in cassava breeding trials, an output that will enhance ongoing efforts to breed for desired end-user root quality traits.Key words: Breeding, cooking, Manihot esculenta
Premise of the study The key to increased cassava production is balancing the trade-off between marketable roots and traits that drive nutrient and water uptake. However, only a small number of protocols have been developed for cassava roots. Here, we introduce a set of new variables and methods to phenotype cassava roots and enhance breeding pipelines. Methods Different cassava genotypes were planted in pot and field conditions under well-watered and drought treatments. We developed cassava shovelomics and used digital imaging of root traits (DIRT) to evaluate geometrical root traits in addition to common traits (e.g., length, number). Results Cassava shovelomics and DIRT were successfully implemented to extract root phenotypes, and a large phenotypic variation for root traits was observed. Significant correlations were found among root traits measured manually and by DIRT. Drought significantly decreased shoot dry weight, total root number, and root length by 84%, 30%, and 25%, respectively. High adventitious root number was associated with increased shoot dry weight ( r = 0.44) under drought. Discussion Our methods allow for high-throughput cassava root phenotyping, which makes a breeding program targeting root traits feasible. We suggest that root number is a breeding target for improved cassava production under drought.
Convolutional neural network (CNN) models have the potential to improve plant disease phenotyping where the standard approach is visual diagnostics requiring specialized training. In scenarios where a CNN is deployed on mobile devices, models are presented with new challenges due to lighting and orientation. It is essential for model assessment to be conducted in real world conditions if such models are to be reliably integrated with computer vision products for plant disease phenotyping. We train a CNN object detection model to identify foliar symptoms of diseases in cassava ( Manihot esculenta Crantz). We then deploy the model in a mobile app and test its performance on mobile images and video of 720 diseased leaflets in an agricultural field in Tanzania. Within each disease category we test two levels of severity of symptoms-mild and pronounced, to assess the model performance for early detection of symptoms. In both severities we see a decrease in performance for real world images and video as measured with the F-1 score. The F-1 score dropped by 32% for pronounced symptoms in real world images (the closest data to the training data) due to a decrease in model recall. If the potential of mobile CNN models are to be realized our data suggest it is crucial to consider tuning recall in order to achieve the desired performance in real world settings. In addition, the varied performance related to different input data (image or video) is an important consideration for design in real world applications.
Portable Vis/NIRS are flexible tools for fast and unbiased analyses of constituents with minimal sample preparation. This study developed calibration models for dry matter content (DMC) and carotenoids in fresh cassava roots using a portable Vis/NIRS system. We examined the effects of eight data pre-treatment combinations on calibration models and assessed calibrations on processed and intact root samples. We compared Vis/NIRS derived-DMC to other phenotyping methods. The results of the study showed that the combination of standard normal variate and de-trend (SNVD) with first derivative calculated on two data points and no smoothing (SNVD+1111) was adequate for a robust model. Calibration performance was higher with processed than the intact root samples for all the traits although intact root models for some traits especially total carotenoid content (TCC) (R2c = 96%, R2cv = 90%, RPD = 3.6 and SECV = 0.63) were sufficient for screening purposes. Using three key quality traits as templates, we developed models with processed fresh root samples. Robust calibrations were established for DMC (R2c = 99%, R2cv = 95%, RPD = 4.5 and SECV = 0.9), TCC (R2c = 99%, R2cv = 91%, RPD = 3.5 and SECV = 2.1) and all Trans β-carotene (ATBC) (R2c = 98%, R2cv = 91%, RPD = 3.5 and SECV = 1.6). Coefficient of determination on independent validation set (R2p) for these traits were also satisfactory for ATBC (91%), TCC (88%) and DMC (80%). Compared to other methods, Vis/NIRS-derived DMC from both intact and processed roots had very high correlation (>0.95) with the ideal oven-drying than from specific gravity method (0.49). There was equally a high correlation (0.94) between the intact and processed Vis/NIRS DMC. Therefore, the portable Vis/NIRS could be employed for the rapid analyses of DMC and quantification of carotenoids in cassava for nutritional and breeding purposes.
Cassava is the third largest source of carbohydrates for human food in the world but is vulnerable to virus diseases, which threaten to destabilize food security in sub-Saharan Africa. Novel methods of cassava disease detection are needed to support improved control which will prevent this crisis. Image recognition offers both a cost effective and scalable technology for disease detection. New deep learning models offer an avenue for this technology to be easily deployed on mobile devices. Using a dataset of cassava disease images taken in the field in Tanzania, we applied transfer learning to train a deep convolutional neural network to identify three diseases and two types of pest damage (or lack thereof). The best trained model accuracies were 98% for brown leaf spot (BLS), 96% for red mite damage (RMD), 95% for green mite damage (GMD), 98% for cassava brown streak disease (CBSD), and 96% for cassava mosaic disease (CMD). The best model achieved an overall accuracy of 93% for data not used in the training process. Our results show that the transfer learning approach for image recognition of field images offers a fast, affordable, and easily deployable strategy for digital plant disease detection.
Background Understanding root traits is a necessary research front for selection of favorable genotypes or cultivation practices. Root and tuber crops having most of their economic potential stored below ground are favorable candidates for such studies. The ability to image and quantify subsurface root structure would allow breeders to classify root traits for rapid selection and allow agronomist the ability to derive effective cultivation practices. In spite of the huge role of Cassava ( Manihot esculenta Crantz), for food security and industrial uses, little progress has been made in understanding the onset and rate of the root-bulking process and the factors that influence it. The objective of this research was to determine the capability of ground penetrating radar (GPR) to predict root-bulking rates through the detection of total root biomass during its growth cycle. Our research provides the first application of GPR for detecting below ground biomass in cassava. Results Through an empirical study, linear regressions were derived to model cassava bulking rates. The linear equations derived suggest that GPR is a suitable measure of root biomass ( r = .79). The regression analysis developed accounts for 63% of the variability in cassava biomass below ground. When modeling is performed at the variety level, it is evident that the variety models for SM 1219-9 and TMS 60444 outperform the HMC-1 variety model (r 2 = .77, .63 and .51 respectively). Conclusions Using current modeling methods, it is possible to predict below ground biomass and estimate root bulking rates for selection of early root bulking in cassava. Results of this approach suggested that the general model was over predicting at early growth stages but became more precise in later root development.
Background Cassava mosaic disease (CMD) is a major constraint to cassava production in sub-Saharan Africa. Under field conditions, evaluation for resistance to CMD takes 12-18 months, often conducted across multiple years and locations under pressure from whitefly-mediated transmission. Under greenhouse or laboratory settings, evaluation for resistance or susceptibility to CMD involves transmission of the causal viruses from an infected source to healthy plants through grafting, or by using Agrobacterium-mediated or biolistic delivery of infectious clones. Following inoculation, visual assessment for CMD symptom development and recovery requires 12-22 weeks. Here we report a rapid screening system for determining resistance and susceptibility to CMD based on virus-induced gene silencing (VIGS) of an endogenous cassava gene. Results A VIGS vector was developed based on an infectious clone of the virulent strain of East African cassava mosaic virus (EACMV-K201). A sequence from the cassava (Manihot esculenta) ortholog of Arabidopsis SPINDLY (SPY) was cloned into the CP position of the DNA-A genomic component and used to inoculate cassava plants by Helios® Gene Gun microparticle bombardment. Silencing of Manihot esculenta SPY (MeSPY) using MeSPY1-VIGS resulted in shoot-tip necrosis followed by death of the whole plant in CMD susceptible cassava plants within 2-4 weeks. CMD resistant cultivars were not affected and remained healthy after challenge with MeSPY1-VIGS. Significantly higher virus titers were detected in CMD-susceptible cassava lines compared to resistant controls and were correlated with a concomitant reduction in MeSPY expression in susceptible plants. Conclusions A rapid VIGS-based screening system was developed for assessing resistance and susceptibility to CMD. The method is space and resource efficient, reducing the time required to perform CMD screening to as little as 2-4 weeks. It can be employed as a high throughput rapid screening system to assess new cassava cultivars and for screening transgenic, gene edited and breeding lines under controlled growth conditions.
Past research developed reliable equations to base selections for high β‐carotene on near‐infrared spectroscopy (NIR) predictions (100 genotypes d −1 ) rather than with high‐performance liquid chromatography (HPLC) (<10 samples d −1 ). During recent harvest, CIAT made selections based on NIR predictions for the first time. This innovation produced valuable information that will help other cassava ( Manihot esculenta Crantz) breeding programs. A total of 284 samples were analyzed with NIR and HPLC for total β‐carotene (TBC) and by the oven method for dry matter content (DMC). Results indicated that NIR reliably predicted TBC and DMC. In addition, 232 genotypes grown in preliminary yield trials (PYTs) were harvested at 8.5 and 10.5 mo after planting (one plant per genotype and age) and root quality traits analyzed (by NIR only). Repeatability of results at the two ages was excellent, suggesting reliable results from NIR. In contrast to previous reports, age of the plant did not influence carotenoids content in the roots. The availability of a high‐throughput NIR protocol allowed comparing results (for the first time) from seedling and cloned plants from the same genotype. Results showed very little relationship for DMC between seedling and cloned plants ( R 2 = 0.09). There was a much better association for TBC ( R 2 = 0.48) between seedling and cloned plants. It is postulated that variation in the environmental conditions when seedling and cloned plants (from the same genotype) may be responsible for these weak associations. Important changes in selection strategies have been implemented to overcome problems related to a lengthy harvesting season.
Abstract Plant disease symptoms exhibit complex spatial and temporal patterns that are challenging to quantify. Image-based phenotyping approaches enable multi-dimensional characterization of host-microbe interactions and are well suited to capture spatial and temporal data that are key to understanding disease progression. We applied image-based methods to investigate cassava bacterial blight, which is caused by the pathogen Xanthomonas axonopodis pv. manihotis ( Xam ). We generated Xam strains in which individual predicted type III effector (T3E) genes were mutated and applied multiple imaging approaches to investigate the role of these proteins in bacterial virulence. Specifically, we quantified bacterial populations, water-soaking disease symptoms, and pathogen spread from the site of inoculation over time for strains with mutations in avrBs2 , xopX , and xopK as compared to wild-type Xam . Δ avrBs2 and Δ xopX both showed reduced growth in planta and delayed spread through the vasculature system of cassava. Δ avrBs2 exhibited reduced water-soaking symptoms at the site of inoculation. In contrast, Δ xopK exhibited enhanced induction of disease symptoms at the site of inoculation but reduced spread through the vasculature. Our results highlight the importance of adopting a multi-pronged approach to plant disease phenotyping to more fully understand the roles of T3Es in virulence. Finally, we demonstrate that the approaches used in this study can be extended to many host-microbe systems and increase the dimensions of phenotype that can be explored. Summary: Novel, image-based phenotyping methods enhance characterization of plant-pathogen interactions.
Plant disease symptoms exhibit complex spatial and temporal patterns that are challenging to quantify. Image-based phenotyping approaches enable multidimensional characterization of host-microbe interactions and are well suited to capture spatial and temporal data that are key to understanding disease progression. We applied image-based methods to investigate cassava bacterial blight, which is caused by the pathogen Xanthomonas axonopodis pv. manihotis (Xam). We generated Xam strains in which individual predicted type III effector (T3E) genes were mutated and applied multiple imaging approaches to investigate the role of these proteins in bacterial virulence. Specifically, we quantified bacterial populations, water-soaking disease symptoms, and pathogen spread from the site of inoculation over time for strains with mutations in avrBs2, xopX, and xopK as compared to wild-type Xam ∆avrBs2 and ∆xopX both showed reduced growth in planta and delayed spread through the vasculature system of cassava. ∆avrBs2 exhibited reduced water-soaking symptoms at the site of inoculation. In contrast, ∆xopK exhibited enhanced induction of disease symptoms at the site of inoculation but reduced spread through the vasculature. Our results highlight the importance of adopting a multipronged approach to plant disease phenotyping to more fully understand the roles of T3Es in virulence. Finally, we demonstrate that the approaches used in this study can be extended to many host-microbe systems and increase the dimensions of phenotype that can be explored.