Rice Leaf Disease Classification Based on Deep Learning

Agriculture is an important foundation of the national economy, and food security is an important guarantee for achieving sustainable development. Rice is one of the main food crops in global food production, and its yield directly affects the quality of life of billions of people. However, in the process of rice cultivation, the occurrence of various diseases poses a serious threat to its yield and quality. According to statistics, the losses caused by diseases can account for 10%-15% of the total rice yield, and even higher in some areas. In order to solve the above problems, this paper mainly studies the application of deep learning technology in the classification of rice leaf diseases, and for this purpose, this paper studies the performance of three mainstream deep learning networks in rice leaf classification, namely ResNet (18, 34, 50), ViT, and Inception-ResNet networks. to enhance the complexity of data features. In this paper, the dataset is divided into training and test sets at a ratio of 7:3, and most of the Acc in the initial test is obtained in this paper, among which Inception-ResNet performs the best, achieving 86.49% and 87.39% accuracy on the original dataset and amplified dataset, respectively, and in the subsequent tests, Inception-ResNet achieves 89.83% results through the combination of Adamw optimizer and LS cross-entropy loss function. This achievement achieves the best results in the experiments designed in this paper. Therefore, based on the model, a WEB platform was developed for visual access to the model using the model weights obtained in training, through which users could quickly obtain the disease status of rice leaves according to the pictures.