Threat Target Detection System for UAV Equipment on the End Side of Transmission Lines

Safety inspection of the transmission line is the essential precondition to ensure the power grid’s safety. The current approach collects images with cameras and then conducts object detection on the cloud. However, due to the insufficient transmission bandwidth, some images may lose and lead to potential risk. This paper proposes a threatening object detection system with a UAV device for transmission lines to address this issue. Our system consists of two steps, first using the YOLOX network to train the model on existing data from the cloud and then implementing the quantized model on the end-side UAV device. The proposed method can detect threatening objects in real-time, which can ensure the safety of transmission lines. Experiments show that the proposed method has better performance on threatening object detection than the state-of-the-art object detection models, including Faster-RCNN, YOLOv3, YOLOv5, etc.