Abstract: With the expanding scale of transmission and distribution networks and the increasing demand for operational safety, conventional manual inspection methods are suffering from low efficiency, high cost, and significant safety risks. To address these challenges, this paper proposes a lightweight vision-based detection and recognition framework for transmission and distribution line. Specifically, a Mobile-YOLOv5 architecture is devised by integrating the MobileNet backbone into the YOLO paradigm, and further optimized through the introduction of an MPDIoU loss function and the Soft-NMS post-processing mechanism, thereby simultaneously improving detection accuracy and computational efficiency. Extensive experiments demonstrate that the proposed system achieves substantial advantages in mean average precision (mAP), inference speed, and model compactness, offering strong technical support for the practical deployment of unmanned-aerial-vehicle-based autonomous inspection systems.