YOLOv5-Based Rice Disease and Pest Recognition Design

This research aims to enhance the speed and accuracy of identifying rice pests and diseases utilizing YOLOv5 object detection algorithm to assisting agricultural producers in effective management. The study focuses on a publicly available data set of images to detect rice pests and diseases. Using YOLOv5 algorithm to train on this data set, developing an effective model that is integrated into a user-friendly interface, allowing producers to quickly identify types of rice pests and diseases by simply uploading images. After noising reduction and coloring transformations of the data set, the YOLOv5 model is trained with the processed images. The model achieved an accuracy and recall rate of 91.3% and 80.6%, respectively. It can accurately identify various types of rice pests and diseases, with all identification accuracies exceeding 86%. Ultimately, the experimental results demonstrate substantial practical value for agricultural producers in daily monitoring and managing rice pests and diseases.