Considering that the most damage detection of conveyor belts is about the tearing of conveyor belt and there is a lacking of other damage types,an improved mine conveyor belt damage detection method is proposed.SPD-Conv module is used to replace the Convolution layer in conv module to improve the detection effect of small targets;Introducing CBAM attention mechanism before backbone feature network and final prediction network to strengthen important feature channels;Finally,Gaussian filter is introduced to eliminate noise interference on the basis of YOLOv5,so as to improve the efficiency of target detection.The experimental results show that the average detection accuracy of the improved YOLOv5 target detection network for four types of damage of conveyor belt is 92.3%,which is 35.1%higher than that of YOLOv5 algorithm,and the detection speed is 90 frames/s,which is 20%higher,thus realizing the rapid identification of mine conveyor belt damage.
维普
万方数据