Transmission line pin detection based on improved SSD

Konferenz: AIIPCC 2022 - The Third International Conference on Artificial Intelligence, Information Processing and Cloud Computing
21.06.2022 - 22.06.2022 in Online

Tagungsband: AIIPCC 2022

Seiten: 6Sprache: EnglischTyp: PDF

Autoren:
Li, Ruihai; Zhang, Wei; Zhang, Guifeng; Yang, Yuxuan (CSG Electric Power Research Institute Co., Ltd., Guangzhou, China)
Yang, Yang (Tianjin University, Tianjin, China)
Li, Nan (Institute of Robotics and Intelligent Manufacturing, The Chinese University of Hong Kong, Shenzhen, China)

Inhalt:
The pin is a critical component of the transmission line, serving a vital role in the most critical areas of the line. If the pin is damaged or falls off, it will cause instability of important components and even a large-scale power outage risk. A new model aimed at the problem of pin defect detection based on the SSD is proposed. Given the small size of the pin, a multibranch feature extraction structure and a residual block are introduced to fully extract the features of the pin; the dilated convolution is introduced to increase the receptive field. For the need of real-time detection in engineering, the convolution split is employed to reduce the running time. The experimental results show that the recall rate of the proposed model in the detection of transmission line pins reaches 87.46%, and the AP reaches 82.22%. This demonstrates that the model has strong expressive power in the identification of pin defects in complex backgrounds and various angles.