Research on Target Detection for Waste Classification Based on Improved YOLOv5

In order to further improve the accurate recognition of targets by a waste sorting robot, this paper proposes a robot target detection algorithm based on improved YOLOv5. By using CoordConv convolution and introducing the SENet attention mechanism, the acquisition and recognition of target information is improved. Through a large number of data comparisons, the improved algorithm proposed in this paper improves the accuracy by 5.3% compared with the traditional algorithm. It can effectively improve the efficiency of rubbish collection and classification.