Development of a deep learning model for the classification and detection of maize leaf diseases

Abstract

Maize leaf blight and maize streak virus are global diseases that significantly damage maize crops and reduce yields. Early detection is crucial for preventing the spread of these diseases and minimizing financial losses. This project aims to develop a deep learning model for automated detection using leaf image data. The proposed method involves collecting images, preprocessing them, extracting features, training the model, and classifying diseases. Researchers from Uganda’s National Crops Resources Research Institute (NaCRRI) and Makerere Artificial Intelligence Lab, both part of the National Agricultural Research Organization (NARO), created the dataset. Feature extraction utilizes convolutional neural networks (CNNs), specifically MobileNet, and a custom model. While the custom model is built from scratch, MobileNet leverages pre-training on large image datasets like ImageNet. To evaluate performance, an extensive dataset of annotated, disease-infected maize leaf images is used, measuring accuracy, precision, recall, and F1-score. Results indicate that the custom-built model for automated detection of maize leaf blight and maize streak virus has high accuracy and shows great potential.