Detection of Bird Nests on Power Transmission Lines Using an Improved YOLOv5s Model with Small Datasets

The safety and reliability of power transmission lines are crucial for the efficient operation of the power system, and the construction of bird nests is one of the common factors threatening the safety of these lines, which may lead to short circuit accidents and power supply interruptions. Therefore, in this paper, we propose a smallsample learning method based on an improved YOLOv5s model to detect bird nests on transmission lines. This method optimizes the backbone network, neck structure, and feature mapping strategy of the model, not only maintaining the same computing speed as the original YOLOv5s, but also averaging the accuracy ([email protected]). The increase of $3 \%$ has effectively improved the safety and stability of the power grid.