IOT – ENABLED EXPLAINABLE AI SYSTEM FOR REAL – TIME CROP DISEASE DETECTION AND SMART FARMING

Abstract

Abstract: Crop diseases significantly reduce agricultural productivity and directly impact farmers’ income. Early detection of plant diseases is essential for effective crop management and the prevention of large-scale crop loss. This paper proposes an IoT-based crop disease detection system using deep learning techniques for automatic identification of plant diseases from leaf images. The system utilizes a Convolutional Neural Network (CNN) model to classify crop diseases from images captured through an IoT camera module [1], [2].To improve interpretability, the Grad-CAM technique is used to generate heatmaps highlighting infected regions on the leaf [4]. Additionally, the system provides automated recommendations for disease management and communicates them to farmers through both GSM-based SMS alerts and a speaker module [11], [16].The speaker module converts prediction results into audio output in the farmer’s native language, announcing the detected disease, recommended pesticide, duration of application, and severity level. This feature enhances accessibility for illiterate and regional-language users. The proposed system integrates image processing, machine learning, IoT devices, and multimodal communication technologies to provide a scalable and practical solution for precision agriculture [9], [10]. Index Terms: Crop Disease Detection, Internet of Things (IoT), Deep Learning, Convolutional Neural Network (CNN), Grad-CAM, Precision Agriculture, GSM Communication, Speaker Module, Audio Output, Smart Farming.

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.