Target Detection of Transmission Line Towers Based on Integrated YOLOv5 Algorithm

Transmission tower is an important component of transmission line, and its safety directly affects the safety and stability of power transmission. According to the low accuracy of small target identification of transmission tower in remote sensing images, in this study, integrated modeling was conducted based on YOLOv5s and YOLOv5x algorithms, weighted boxes fusion (WBF) reasoning mechanism was added, model training was conducted with high-resolution remote sensing tower image data set, and performed a test-time augmentation of the data set. The experimental results showed that compared with the single model recognition results, the recognition accuracy, recall rate and mAP@.5 of integrated YOLOv5 model are significantly improved, reaching 0.952, 0.944 and 0.929 respectively. In addition, under some complex background, different illumination environment and different weather conditions, the model proposed in this paper has good recognition effect and strong robustness.