Surface waste identification and detection based on improved YOLOV8

In response to the challenge of surface waste detection, an advanced surface waste recognition algorithm based on YOLOV8 has been carefully developed. The RepVGG module is used to replace the traditional Conv module, so the detection speed and accuracy of the model are significantly improved in the inference stage. At the same time, in order to further enhance the training effect of the model, Wasserstein Distance Loss was introduced as a loss function to optimize the training process of the model and further enhance its training accuracy. After rigorous experimental verification, the experimental data have excellent performance, and the improved network model has an average precision mean (mAP) of 82.6%, which proves its high-precision performance in target detection, effectively reduces the case of missing detection, and meets the strict requirements for surface garbage detection.