Mobile-Driven Disease Identification System for Rice Plants Using ResNet

Abstract

Rice is one of the staple food sources of the Indonesian people. Rice is one of the staple food sources of the Indonesian people. However, rice production in Indonesia has decreased significantly. This has led to more rice imports in Indonesia. It makes Indonesia more dependent on other countries. The main factor in the crop failure of rice is disease. Agricultural experts are still lacking in Indonesia. This makes the problem of crop failure persist. Therefore, we developed a mobile-based system to identify rice plant diseases. It provides treatment advice for infected plants. This system works in reference to one of the artificial intelligence methods, namely convolutional neural networks. The system is programmed to learn and recognize interconnected networks that form a pattern. So that it can understand similar patterns in different images. In this study, we used the ResNet-50 model with Adamax optimization. It got a training accuracy of 99.94%. To use this application, users only need a smartphone and do not need any internet access. The app has high mobility and easy access for farmers during fieldwork activities. It can solve the problem of crop failure. It's especially helpful in rice fields.