Smart multi-purpose farm disease monitoring and notification model

Abstract

Disease has remained a threat to the existence of every living thing on earth. Particularly in agriculture, disease has remained a major constraint to the success of crop yield and demands urgent solutions, starting with early detection. While many studies have been presented on plant disease detection and control, despite their success, the research gap resides in creating a balance between plant disease detection and farm disease detection. This is because detecting disease in a plant does not necessarily imply that the farm is infected with diseases, and this has resulted in issue of false alarm in the existing system. To address this challenge, YOLOV-5 model was trained with a plant disease dataset considering diverse classes of plants such as corn, waterleaf, tomato, pepper, and cassava. The plant disease model generated was used to develop a farm disease monitoring, detection and notification algorithm, which was converted into mobile application software using Python programming for real-time monitoring notification of farm diseases. This multi-purpose system when tested, reported an average precision mean of 0.95. In addition, experimental validation of the model in maize and watermelon farms reported real-time disease detection and notification which facilitates rapid response and control of the disease by the farmer.