Vietnam is a significant exporter of agricultural products in the region and the world. However, farmers face the impact of global climate change, which creates conditions for many types of pests through stormy and sunny weather, where rice diseases can develop rapidly. Therefore, early detection and timely treatment of rice diseases are paramount to farmers. Therefore, there is an urgent need to find a method that can quickly and accurately distinguish multiple rice disease images. Machine learning-based methods can be a potential solution for image classification for rice disease detection. They can process a large amount of data and increase the accuracy of disease diagnosis so that farmers can detect the disease to provide treatment solutions promptly. This paper proposes a solution for disease detection in rice leaves using three transfer learning models, including EfficientNetB3, VGG-16, and MobileNetv2. The proposed method achieved 90%, 93%, and 94% accuracy, respectively, in detecting nine types of diseases and normal leaves. These results can be used to predict diseases on rice leaves from images, suggesting appropriate prevention and treatment solutions to help farmers improve rice productivity.