Performance Analysis of CNN Architecture in Rice Leaf Disease Image Classification

The decline in rice yields in Indonesia is caused by the attacks of plant pests (OPT), including leaf diseases such as leaf blast, bacterial blight, and dwarfing. Early identification of rice diseases is a crucial step to improving productivity and food security. This study utilizes Deep Learning with Convolutional Neural Network (CNN) to detect rice leaf diseases based on digital images. The dataset consists of four classes of rice leaf diseases, totaling 5,932 images. The research process involves data collection, preprocessing, data augmentation, model training, and evaluation using accuracy, precision, recall, and F1-score metrics. The results indicate that the CNN model achieved the highest accuracy of 99% on validation data with a training-to-validation data ratio of 80:20. The evaluation using a confusion matrix demonstrates excellent performance in distinguishing disease types. With high accuracy levels, this model has the potential to become an effective tool for early detection of rice leaf diseases, providing practical solutions for farmers in managing plant diseases.