Track installation operations in high-altitude transmission systems are often challenged by attitude deviations, structural misalignments, and other anomalies that pose serious safety risks. Traditional manual inspection methods are time-consuming and prone to errors, necessitating more accurate and automated detection solutions. This study aims to enhance the precision and robustness of anti-fall track installation detection by integrating improved deep learning techniques with UAV image acquisition. To achieve this, an image processing strategy based on the AugMix algorithm is employed to generate diverse and noise-resistant samples. A modified YOLOv7 framework incorporating the Swin Transformer module is developed to improve detection accuracy and efficiency. Experimental results demonstrated that the proposed model achieved a Top-1 accuracy of 88.90%, a signal-to-noise ratio of 0.99, and a structural similarity index of 0.91 in image preprocessing. For the AFTI detection task, the enhanced model attained an accuracy of 0.92, an F1-score of 0.88, and an inference time of 2.17 s in standard conditions, and maintained 0.85 accuracy in complex task scenarios. These results confirm that the proposed approach significantly improves detection accuracy, robustness, and real-time performance, making it a practical solution for automated structural inspection in UAV-based high-altitude operations.
Che et al. (Mon,) studied this question.