Garbage Classification Using YOLO: Model Comparison and Future Directions

This paper explores the application of YOLO-based deep learning algorithms in garbage classification. Models Yolov3 (Darknet53) and Yolov5s/m (CSP-Darknet53) are trained on three recyclable and three non-recyclable categories, comparing their prediction accuracy and speed. Results indicate that Yolov5s outperforms Yolov3 in recognition rate and efficiency, especially on resource-limited devices. The study highlights the potential of YOLO algorithms in intelligent garbage sorting and suggests future improvements for this application.