Vision-Based Personnel Positioning Framework in Power Operation Environments

Cost and reliability are the primary challenges for personnel positioning in the power operation environment. Mainstream base station positioning methods, such as Beidou and UWB, are hindered by high costs and limited positioning accuracy due to strong electromagnetic interference in power scenarios, leading to significant errors. To address the aforementioned issues, this paper, based on the current context of digital substation construction, proposes a two-stage visual personnel positioning method that utilizes existing video surveillance systems to identify key personnel nodes and estimate camera pose. Additionally, to tackle challenges of occlusion and jitter in complex environmental positioning, the paper optimizes the approach by using multiple object detection models and Kalman filtering, with a positioning demonstration conducted on a real collected video.