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

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].