In many Asian countries, rice plants provide essential resources of food. Quite a number of people depend on rice products in daily life. However, diseases have caused a critical damage for production of rice plants every year. Therefore, the early diagnosis and treatment of rice diseases become important. The automatic and accurate classification of rice diseases are crucial for the treatment purposes. In the past decade, deep neural networks (DNNs) have achieved a great advance in classifying plant diseases. The paper introduces a multiclass classification approach for identifying diseased rice leaves using data augmentation technique and DNNs. First, we apply data augmentation strategy using traditional image processing to increase the number of images in the original dataset. Next, the Vision Transformer (ViT) network is applied and optimized to enhance the accuracy of classifying diseases in rice plant leaves. The proposed method is evaluated on two public datasets of diseased rice plant leaf images. The proposed method achieved a classification accuracy of 92.5% of rice diseases on two large datasets. Moreover, performance comparisons with various existing methods have shown the outstanding and promising applications of the proposed method.