Rice Leaf Diseases Detection Using YOLOv8

Abstract

The development of rice plants holds immense importance today as it impacts crucial aspects of life such as food security, agricultural advancement, and the economy of nations. Consequently, research on disease detection in rice plants, particularly using machine learning, is gaining popularity. Several diseases pose a threat to rice leaves, with Leaf Blast, Leaf Folder, and Brown Spot being the most common ones, directly affecting crop cultivation and causing yield loss. In this study, we propose the utilization of deep learning, the state-of-the-art image processing solution, to address this issue. Our proposed method consists of two steps: first, collecting reliable dataset by approaching and capturing direct images of rice leaf diseases in the fields, and second, designing and training an Artificial Intelligence (AI) model using the YOLOv8 algorithm to detect and classify the three aforementioned diseases. The data set used in this study includes 3175 images, divided into three parts, of which the training part is 2608 images, the validation part is 326 images and the test part is 241 images. Our experimental results demonstrate an accuracy up to 88.9% for the proposed model.