CT-YOLOv7: Enhancing YOLOv7 for Bird Nest Detection on Power Transmission Lines

Electricity is an important resource indispensable for the development of modern society, and it is also a public facility related to the national economy and people's livelihood. Maintaining a stable supply of electricity is the key to the stable development of national and social production.In order to ensure the stable operation of transmission lines, it is necessary to monitor the nesting behavior of birds on overhead line towers. In order to solve the bird's nest detection task of transmission lines, a CT-YOLOv7 network is proposed based on the problems that the basic YOLOv7 has insufficient feature extraction capabilities and small receptive field in the bird's nest detection task. First, a receptive field block is integrated into the backbone to increase the receptive field for feature extraction. In addition, the head network of YOLOv7 enters the context transformer attention mechanism to extract key feature information and integrate the information of image modality and semantic modality to improve model detection. ability. The CT-YOLOv7 network proposed in this article was applied to the self-made data set for experiments. The results showed that the mAP of the improved YOLOv7 detection model reached 97.3%, which was 3.2% higher than the original YOLOv7 model. Therefore, using the CT-YOLOv7 proposed in this article to detect bird's nests in transmission lines can improve the accuracy, reduce transmission line faults such as insulator flashovers and phase-to-phase short circuits caused by bird damage, and improve inspection efficiency.