The quantitative evaluation framework for non-structural hazards to transmission tower caused by bird nest using large images obtained by unmanned aerial vehicles

Bird nest, as a major foreign body on high-voltage transmission lines, has potential hazards to the safety of transmission lines. Its branches may burn in hot and dry weather, and fall between insulators as conductors in rainy weather, resulting in insulator flashover. Therefore, it is of great significance to detect the bird nest's positions on the transmission line in time and eliminate the hidden danger. This paper proposes a bird nest detection framework based on the panoramic image of the transmission tower. Firstly, the transmission tower is located and Simple based on Swin Transformer is used Baseline detected its 3D attitude to obtain its key points, then clipped its key points and used DenseNet to judge whether the key points have nests, so as to realize the detection of tiny nests in the large-resolution power line vista image and judge the position relationship between them and the transmission tower, so as to realize the automatic inspection of the nests of transmission lines. Experimental results show that the proposed framework can achieve 86.15%,76.71%,81.16% accuracy, recall rate and F1 score in self-made bird nest and pylons data sets.