Rice Leaves Disease Detection and Classification Using Transfer Learning Technique
Rukhsar, Santosh Kumar Upadhyay
Abstract
The demand for food grains is increasing as the world's population grows. Half of the country's population relies on paddy as a food source. Most farmers struggle with the challenges in the early detection of rice leaf disease. Rice leaf diseases such as Blast, blight, and tungro can be detected early using transfer learning models, allowing us to prevent disease spread over the entire plant and thus increasing the rice yields. Using transfer learning techniques DenseNet201, the suggested system has been used to identify diseases in rice crops. This research focuses on three well-known rice diseases: fungus-caused blast, bacteria-caused bacterial leaf blight, and virus-caused tungro. A total dataset of 240 images of three classes is taken. When compared to existing models, our experimental results analysis of DenseN et201 has a higher accuracy of 96.09 percent. The same data set is also applied to simple CNN, which has achieved a 62.20% accuracy.