Safety monitoring method for powerline corridors based on single-stage detector and visual matching

Abstract

Effective monitoring and early warning of overhead power lines are critical for ensuring power supply stability and personnel safety. However, existing methods that rely solely on single vision or light detection and ranging (LiDAR) have limitations such as numerous invalid alarms, hardware equipment constraints, and lack of real-time monitoring capabilities. This article proposes an intelligent monitoring method for safety distance of the powerline corridor based on heterogeneous sensor information. Firstly, an optimised single-stage detector is designed to detect safety hazard objects. In the detector, the backbone network and the sample matching strategy are improved; in addition, the learning strategy is adjusted. Next, the pose transformation relationship is obtained through the visual matching of prior LiDAR information and images. The back projection transformation of 2D–3D is achieved according to the relationship. Finally, a monocular camera-based end-to-end distance measurement scheme is proposed by combining 2D object information with coordinate transformation relationships. The scheme is applied to the distance measurements from hazard objects to powerlines. The experimental results show that the optimisation method improves the detection accuracy and reduces the computational complexity of the model. Also, case experiments with continuous frame data verify the effectiveness of the safety distance monitoring scheme.