Intelligent Leakage Inspection Method of Magnetic Mill Workshop based on Lightweight Yolov5

In the process of mining company beneficiation, large rotating equipment, pipelines and electrical equipment have many complex inspection areas and diverse inspection contents. The lack of intelligent solutions to the problem of pipeline leakage will cause serious consequences such as equipment shutdown if it is not found in time. This paper proposes a lightweight yolov5 intelligent leakage inspection method for the magnetic mill workshop. The inspection robot is used to continuously collect images of possible leakage points and transmit them to the background server. The improved yolov5 algorithm simplifies the model structure and realizes rapid and accurate identification of leakage targets with limited computing resources. The computational efficiency is increased by 44.98% with almost no loss of model accuracy. The mAP_0.5 of identifying large targets at leakage points reaches 97.96%.