Rice Leaf Disease Detection Using ResNet-50 Based Deep Convolutional Neural Network

Abstract

Rice leaf diseases remain a major challenge to crop productivity, particularly in regions where rice serves as a primary staple food. Early and accurate identification of leaf diseases is therefore essential to reduce yield loss and support timely treatment. This study proposes a deep learning-based approach for automatic rice leaf disease classification using the ResNet-50 architecture. The proposed framework is designed as an end-to-end multiclass classification model for six rice leaf conditions, namely Bacterial Leaf Blight, Brown Spot, Healthy Rice Leaf, Leaf Blast, Leaf Scald, and Sheath Blight. The dataset used in this study consists of 3,197 rice leaf images with a relatively balanced class distribution. The main contribution of this work lies in developing a practical image-based disease classification framework that leverages ResNet-50 for reliable feature extraction and multiclass prediction in agricultural settings. Experimental results show that the proposed model achieved an accuracy of 85.08%, precision of 85.76%, recall of 85.08%, and F1-score of 85.07%. These results demonstrate that the proposed framework is effective for multiclass rice leaf disease classification and has strong potential to support automated plant disease detection in intelligent agriculture and precision farming systems.