Image-Based Deep Learning Method for Detecting Diseases in Rice Plants

Abstract

India's rapidly growing population demands increased agricultural productivity, particularly in rice cultivation, which is vital to national food security. However, rice crops are highly susceptible to various leaf diseases such as Brown Spot, Blast, and Bacterial Blight, which significantly reduce yield and quality. Timely and accurate identification of these diseases is essential to prevent widespread outbreaks and reduce excessive pesticide use. This study proposes a deep learning-based approach for automated rice leaf disease detection. Three state-of-the-art models—YOLOv5, DenseNet-201, and ResNet-101—were evaluated using a publicly available rice disease dataset. All models were trained and tested on the Google Colab platform. This study aimed to identify the most effective and practical model for real-world deployment. The results show that YOLOv5 outperformed DenseNet-201 and ResNet-101 in detection accuracy, offering a robust and scalable solution for real-time disease identification. This work contributes to precision agriculture by enabling early diagnosis and informed decision-making, ultimately improving crop health and promoting sustainable farming practices.