Detecting Sugarcane Pests and Diseases Using CNNs for Precision Crop Detection and Management

Abstract

Effective disease management is essential for sustaining sugarcane yield and quality, and traditional methods, such as visual inspection and chemical analysis, are often costly and time-consuming. This study proposes an innovative solution that leverages artificial intelligence (AI) through Convolutional Neural Networks (CNNs) for advanced crop detection and management in sugarcane farming. The AI precision system aims to automate the detection of sugarcane pests and diseases by analyzing collected imagery using machine learning algorithms. This system processes various image parameters, including leaf color, pest types, damage areas, and texture, to accurately identify diseases. By integrating machine learning with image processing techniques, the system provides farmers with rapid and precise diagnoses, enhancing their ability to manage crops effectively and efficiently. The study employs a descriptive developmental approach, emphasizing the AI-driven system's design, development, and evaluation. The objectives include creating a CNN-based system to capture and analyze sugarcane imagery, facilitating timely pest and disease detection, and assessing the system's quality and usability. Integrating AI in sugarcane agriculture promises significant advancements in crop monitoring and management. This research aims to contribute to precision agriculture, enabling farmers to optimize sugarcane cultivation, reduce losses, and enhance productivity. The proposed system offers a scalable solution for real-time monitoring of extensive crop fields, promoting sustainable agricultural practices and supporting economic stability.

Code and data availability

The paper describes a sugarcane pest/disease image dataset (3,895 images) uploaded to Roboflow and a CNN model, but provides no public deposit, repository URL, DOI, or availability statement for the dataset, trained model, or code. Roboflow is used as a service, not cited with an authors' public asset URL. No paper-fig

No evidence-backed public reproduction asset is currently recorded.