A mobile, AI-Enabled platform for screening of bean plant diseases

Abstract

Beans are one of the most grown and most consumed crops in the world and in over 90% in Uganda. In order to solve the problems faced by the existing system, a computer vision based system is proposed for the classification and early detection of diseases on bean crops using image processing and sending disease information to the farmer using mobile computing. A deep learning approach is proposed to identify and classify beans leaf disease by using a custom dataset of leaf images collected from various districts in Uganda as well as object detection for localization of disease spots on the surface of the bean leaves of 2 specific diseases; Angular leaf spot and Bean Rust as the most common diseases affecting beans in Uganda. these greatly affect the crop yields at the end of every season leading to great losses in terms of food and monetary terms to these farmers. eradicating these diseases at an early stage would greatly improve the bean plant production in the country for greater export and income to the farmers. in this project, we trained image classification models and object detection models and obtained our best results of 97% accuracy, 98% precision and 98% recall these models are deployed on a mobile application to enable easy implementation by the farmers.