Lightweight magnetic leakage signal detection of pipelines based on YOLOv5
Oil and gas are mainly transported through pipelines,while oil and gas pipelines may cause defects due to various internal and external factors.In order to accurately locate the distribution of pipeline defects and facilitate deployment at the edge,the YOLOv5 network was improved in this paper.Firstly,in order to reduce the number of network parameters,the backbone network of YOLOv5 is replaced by the more lightweight MobileNetv3,and the number of channels of partial convolution is optimized.Then,in order to make full use of information such as vector Angle between the real box and the predicted box,SIoU is used as the loss function of YOLOv5.Finally,the CARAFE up-sampling operator is introduced into the network to obtain a large receptive field.The results show that compared with the original algorithm,the parameters of the improved algorithm are reduced by 79.8%,the model size is reduced by 77.4%,and the mAP is increased by 0.8%,which improves the detection accuracy as well as the lightweight.
oil and gas pipelinesmagnetic leakage signalYOLOv5Lightweight
万方数据
维普
