Detection Method of Transmission Line Fittings Based on Improved YOLOv5s

Aiming at the complex background of aerial images of transmission lines collected by UAV, the number of objects to be detected is large, and there are some problems such as false detection and missed detection in the detection process caused by partial occlusion overlap. This paper proposes a detection method of transmission line fittings based on improved YOLOv5s. In this paper, the border feature extraction module is introduced from the border feature, and the border feature extraction operator is used to adaptively extract the border feature from the limit point of the border to enhance the existing point feature. and the Convolutional Block Attention Module (CBAM) is introduced. In order to enhance the attention to the gathered objects, effectively solve the problem of missed detection and false detection caused by occlusion and overlap between fittings. the experimental results show, Compared with the YOLOv5 s algorithm, the mAP of the improved algorithm is increased by 2.1 %. It has higher detection accuracy while ensuring higher detection rate, and has certain application value.