Research Article
Improved YOLOX Transmission Line Insulator Identification
@INPROCEEDINGS{10.1007/978-3-031-31733-0_17, author={Zhongqi Zhao and Qing He and Sixuan Dai and Qiongshuang Tang}, title={Improved YOLOX Transmission Line Insulator Identification}, proceedings={Smart Grid and Innovative Frontiers in Telecommunications. 7th EAI International Conference, SmartGIFT 2022, Changsha, China, December 10-12, 2022, Proceedings}, proceedings_a={SMARTGIFT}, year={2023}, month={5}, keywords={Insulator identification Target detection Rotating box detection Feature pyramid pooling}, doi={10.1007/978-3-031-31733-0_17} }- Zhongqi Zhao
Qing He
Sixuan Dai
Qiongshuang Tang
Year: 2023
Improved YOLOX Transmission Line Insulator Identification
SMARTGIFT
Springer
DOI: 10.1007/978-3-031-31733-0_17
Abstract
Aiming at the problem of low recognition accuracy of insulators in power system transmission lines and the recognition results contain many backgrounds, this paper proposes a high-performance detection model by combining the improved YOLOX target detection algorithm with the rotating frame detection algorithm. Firstly, the backbone network of YOLOX is replaced with ConvNext with a larger receptive field to improve the feature learning ability of the model for insulators. Secondly, the fusion between the output features of the feature pyramid pooling module is enhanced using the channel disorder operation. Finally, the angle classification of the detection frame is added to the network to realize the rotation frame detection and reduce the background interference in the recognition result. The model is trained and tested with manually marked aerial photography data. The test results show that the method has high accuracy in insulator identification and meets the high-performance detection requirements.
