In order to solve the problem of on-line real-time detection to the mine steel wire rope defect and further improve the flexibility and accuracy of the mine steel wire rope defect detection, the research on the image-based mine steel wire rope defect detection was carried out. The camera is used to sample the wire rope before the steel wire rope enters the mine, and a small defect detection model on the surface of the object based on YOLO v5 is proposed, which can accurately detect small defects outside the wire rope. The transfer learning method is introduced to further improve the model accuracy of small sample training. A large number of tests show that the average accuracy of the model are obviously improved compared with unmodified models, and the detection speed can be kept at the real-time level.
文章内容来自网络，如有侵权，联系删除、联系电话：023-85238885
参与评论
请回复有价值的信息，无意义的评论将很快被删除，账号将被禁止发言。
评论区