Classification of Rice Leaf Disease Images Using Convolutional Neural Network (CNN) Algorithm

Abstract

Rice (Oryza sativa) is a major commodity that plays an important role in Indonesia's food security. However, rice productivity often decreases due to leaf diseases, such as neck blast, leaf blight, and rice leafhopper. Manual disease identification still has limitations, as it requires a long time, depends on farmers' expertise, and may lead to misclassification. To address these issues, this study develops a rice leaf disease classification system using a Convolutional Neural Network (CNN) algorithm. The dataset used was from Kaggle, consisting of a total of 3,631 images divided into three disease classes. The data was split with a ratio of 80% for training, 10% for validation, and 10% for testing. The pre-processing steps included resizing, augmentation, and image normalization. The CNN architecture was custom-built with several convolutional, pooling, flatten, and dense layers. The training results showed that the model could achieve a training accuracy of 97.80% and a validation accuracy of 97.42%. The model was then implemented into a web application based on Flask, allowing users to upload images of rice leaves and obtain classification results quickly, accurately, and in real-time. Based on the research results, CNN has been proven effective in classifying rice leaf diseases with a high level of accuracy. This system is expected to help farmers detect diseases early, reduce the risk of crop failure, and support the implementation of smart farming in Indonesia.