A guided wave damage identification method based on the deep learning target detection algorithm is proposed. According to the wave number variation characteristics at the local damage of the structure, the image recognition algorithm is used to detect the full-field wave field images of the structure, thereby achieving damage location and identification. In the process of acquiring training image samples, a series of numerical models of aluminum plates with blind-hole damages at different positions are established, and through multi-frequency excitation, the steady-state wave field images of the structure are obtained, and the sample database is expanded by means of image enhancement technology. YOLOv5s network is selected for model training, and detection is performed on the time-domain guided wave fields of the simulation model and the experimental structure respectively. The results show that when the guided wave propagates through the damaged area, the guided wave field exhibits local distortion at the damage location, and the damage detection box is consistent with the actual damage characteristics of the structure, therefore, the target detection algorithm can avoid the image features of the excitation points and effectively capture the image features of blind-hole damages.