Based on a small number of low-resolution transmission line samples from the Huma Mountain experimental laboratory of Yunnan Power Grid and predecessors, using the super-resolution algorithm, a data augmentation algorithm based on dynamic background and random transformation is proposed, which generates 7 000 clear transmission line images, and solves the data diversity problem of random target positions, random angles and random sizes. Meanwhile, YOLOv5 target detection algorithm is developed to achieve real-time and robust detection of transmission lines, loose strands and broken strands, and reach an average F1 score of 94.7%. The proposed super-resolution and data augmentation method can be widely used in low-resolution images and small sample datasets in various fields. The proposed transmission line and its defect detection algorithm can be applied to the field of transmission line inspection, which make inspection more efficient and intelligent.