AI-Powered Plant Disease Diagnosis via Mobile App for Smarter Farming

Abstract

Plant diseases significantly reduce the quality and quantity of food, fiber, and biofuel crops, posing a serious challenge as agriculture strives to meet the demands of a rapidly growing global population. Both commercial farmers and hobbyist gardeners face increasing concerns about the health of their plants due to these diseases. Moreover, many regions lack access to agricultural testing facilities or experts who can diagnose and treat plant diseases. To address this issue, we propose a solution that allows users with smartphones and WhatsApp to take images of affected plant areas and send them to a WhatsApp Bot. The bot responds with segmented images highlighting the diseased regions, alongside the disease type and its probability. This system utilizes HSV Filtering, MobilenetV2, and YoloV8 algorithms for disease detection, providing accurate and reliable results. Instance segmentation is implemented to identify multiple diseases on a single plant part. With an accuracy of 95.66% for the classification model and a MAP of 0.756 for segmentation, this system enables easy and accessible plant disease diagnosis. By eliminating the need for physical visits to agricultural centers, this solution promotes early detection and treatment of plant diseases at no extra cost, offering significant convenience to farmers.