Image recognition of garbage classification based on YOLOv8

Abstract

In order to realize the rapid and accurate intelligent garbage sorting of robots, the research will build an image recognition model for garbage classification and analyze it. In this study, WP-UT2000 color camera will be used as the image acquisition equipment, and YOLOv8 image recognition model will be used for image recognition of garbage classification. SIoU loss function of angle will be introduced to improve YOLOv8 image recognition model, and the image recognition model will be trained through different data sets to realize robot garbage sorting. The improved YOLOv8 image recognition model has faster convergence speed and higher mAP value. Compared with other loss functions, the SIoU loss function used in the improved YOLOv8 image recognition model has higher accuracy of garbage image recognition. The improved YOLOv8 image recognition model constructed in this study can realize fast and accurate garbage image recognition, and has outstanding practical value in the field of robot garbage sorting.