Classification of rice leaf disease with cutting-edge deep learning models

Global food security depends on rice as a main crop, yet several leaf diseases endanger rice production. To preserve productivity and quality, early and efficient disease identification is essential. Accurate categorization and early diagnosis of rice leaf diseases can now be achieved more quickly because of recent deep learning and image processing automation developments. This study presents a novel method for classifying rice leaf diseases using the latest Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs). We examine the effectiveness of various deep learning models, such as a customized 5-layer CNN, AlexNet, ResNet, ViT, MobileNetV2, and InceptionV3, using a dataset including 5,932 photos of rice leaves classified into four disease categories: brown spot, bacterial blight, blast, and explosion. Metrics like accuracy, precision, recall, and F1-score were used to assess the models. With an F1 score of 0.99 and the highest accuracy of 99% among the models, MobileNetV2 outperformed earlier benchmark results. Our results show that MobileNetV2 performs better for rice leaf disease classification due to its good feature extraction capabilities and efficient architecture. This work supports applying effective models for real-time disease detection in crop management systems and emphasizes the potential of sophisticated deep-learning models to improve agricultural operations.