EpNet: Power lines foreign object detection with Edge Proposal Network and data composition

Junyu Su,Yukun Su,Yu Zhang,Weiqiang Yang,Huichou Huang,Qingyao Wu

DOI: https://doi.org/10.1016/j.knosys.2022.108857

2022-04-01

Abstract:Power lines foreign object detection task is to detect objects suspending on power lines, they might be kites, plastic bags or anything else, which could be a potential risk to power system. However, this task remains a challenge due to the lack of data, because these data can only be produced by a few major video surveillance companies, who treat their data as valuable property and won't share it with others. Without massive training data, we could not obtain an excellent neural network. In this paper, we introduce a new data composition method to generate artificial data and help alleviate the data shortage problem. What's more, we propose a new detection method called Edge Proposal Network (EpNet) to reduce wrong proposal locations and increase detection performance. At last, we conduct several experiments to verify the effectiveness of the two methods, and some discussion experiments to gain a deeper understanding of the composited data.

computer science, artificial intelligence