Research on garbage classification model based on multi-scale feature extraction and fusion
Taking the YOLOv4 algorithm as the foundation,the paper proposes improvement strategies for the YOLOv4 algorithm by understanding its basic structure and addressing the issues encountered in practical applications of garbage classification.The improvement is achieved by setting two convolutional kernels in the CBM module to facilitate the feature extraction process for garbage of different sizes.Through training and validation of the improved garbage classification model with multi-scale feature extraction and fusion,it is found that the improved YOLOv4 algorithm can effectively meet the basic requirements of multi-scale feature extraction and fusion.The recognition efficiency for small-sized garbage targets can reach 85%over,and the processing time can meet the basic requirements of practical applications.Furthermore,it is observed that as the convolutional kernel size decreases,the accuracy of garbage target identification increases while the processing time also increases accordingly.This research contributes to the advancement of the garbage classification industry towards intelligence.
万方数据
维普
