Smart detection and classification of fungal disease in rice plants using image processing techniques

Abstract

One of the most crucial staple crops, rice (Oryza Sativa), feeds a significant percentage of the world’s population. However, fungal infections, which may significantly reduce yields and affect global food security, represent an extreme risk to rice’s productivity and quality. We created a custom dataset of 991 images capturing both healthy and False smut affected rice plants. Several state-of-art deep learning models including ResNet50V2, AlexNet, VGG19, VGG16, InceptionV3, and CNN architecture were applied to classify the disease. The models were trained, validated and tested on our dataset, and the performance was analyzed based on metrics such as accuracy, precision, recall, and F1-score. Among all the models, Inception V3 achieved the highest result with an accuracy of 99.49%. The result of the research will further contribute to developing a web application for identifying and diagnosing fungal blasts in rice plants to ensure better rice cultivation, enabling early intervention and sustainable crop management practices.