Key Part Detection of Transmission Line Fitting based on Keypoint Regression
Yanwu Dong, Ziying Lu, Ziqiang Lu, Ruikai Zhu, Chao Dong, Kai Sun, Chun Wang, Juan Du · 2021
The identification of metal defects in transmission lines can reduce the probability of power grid operation failure, which is very important for the stable operation of power grids. The location information of the key parts of fitting is helpful for target location and defect identification, especially defects caused by incorrect or incomplete key connections of parts. However, little research has been done on the location information of key parts of fittings. First, we combine the structure of the strain clamp and the function of key components, design keypoints of the key parts of the strain clamp, and make it into a standard format data set. Then, we design a keypoint feature extraction network based on the convolution layer according to the keypoint dataset that we make for strain clamp. Finally, we use HRNet32 and other networks to perform comparative experiments to prove the effectiveness of our method. Moreover, through the visual analysis of the results, the keypoints of the fittings can be well detected in the complex background or in the case of partial occlusion.