ABSTRACT
Piping is the external manifestation of defects in traditional Chinese medicines in dams. Preventive detection, investigation, and repair are required to avoid water conservancy safety accidents. In view of the reality of the long cycle, low efficiency, and limited detection methods, which are mainly used in the inspection of piping, there are also problems such as missed inspections and the inability to comprehensively analyze the potential risks of piping. After analyzing a variety of piping detection methods, this paper proposes a thermal infrared temperature difference piping detection method based on deep learning and computer vision and uses YOLOv3 to achieve rapid detection of piping infrared images. By analyzing the detection results, an evaluation model for the danger of piping is established, which provides a new technical solution for the detection of piping.
