Abstract: In the tension stringing process, the ability of human visual resolution makes it difficult to recognize details of fast-moving objects, which means that quality problems such as broken wires in fast-moving wires in tension machines cannot be detected in time. This leads to a large amount of work in defect elimination for the subsequent high-altitude operations, and even poses safety hazards due to wire quality problems. To address this issue, a defect database for wires is established, and the target detection video recognition algorithm is optimized to form a wire defect recognition model with high recognition rate and low false alarm rate. An intelligent recognition system for wire defects of tensioning machine in transmission line erection is designed and developed, which uses cameras to intelligently identify wire defects such as broken wires and timely remind the commanding officers to handle or mark them. This system reduces the amount of high-altitude work required to handle wire defects, and has achieved good results in practical engineering applications.