Research on The Detection Method for Insulation Piercing Connectors and Bolts on The Transmission Line Based on SSD Algorithm

Aiming at the problem that the piercing clamp and bolt on transmission line is easily effected by illumination, occlusion, environmental background and shooting angle, a method for detecting the piercing clamp and bolt on transmission line based on SSD algorithm is proposed in this paper. In this method, the data set get enhanced by means of flipping, translation and angle rotation, The method of multi-scale feature fusion is used to fuse deeper features of shallow feature maps, increase the perception field of shallow feature maps and improve the detection accuracy for small objects, and the intrinsic VGGNet Network in SSD Architecture is replaced with the Deep Residual Networks(ResNet-50) with deeper network depth and less computational complexity to extract the image feature. The result turns out that the mAP value of SSD models with ResNet50 reaches 91.5%, which increases by 3.2% compared with that of the models using VGG16. By using the Deep Learning object detection model, the puncture clamp with different resolution and different position angles can be better detected, which has high practical value for engineering application.