Rice is a staple food source for most people around the world, including Indonesia, which is an agrarian country where most of its population grows and consumes rice. However, rice plants also suffer from various diseases, especially on the leaves, such as bacterial leaf blight, rice blast, dan rice tungro. If the infection or disease in rice plants is not identified early on, it will decrease production and harm farmers. To address this problem, information technology can be utilized in identifying diseases using image processing and image classification. The dataset was taken from public repositories, and data augmentation was also used in this research to increase the dataset's training accuracy. With this background, a disease detection system approach is proposed using Deep Learning method using Convolutional Neural Network (CNN) and several transfer learning architectures, namely VGG16, NASNetMobile, and Xception, for rice leaf disease detection. The best experimental results were obtained using the Xception architecture, where the training accuracy value is 99.13%, validation accuracy is 97.22%, and the testing accuracy is 97.22%.