Welded joint TOFD detection defect automatic identification method based on deep learning

The invention discloses a welding joint TOFD detection defect automatic identification method based on deep learning, and relates to the technical field of ultrasonic non-destructive detection.The method is characterized in that defect features of an image and an original waveform dimension are extracted and fused by constructing a typical defect TOFD detection map sample library; the method comprises the following steps of: respectively combining with a single-stage target detection algorithm and a generative adversarial mechanism to construct an abnormal welding seam classification model and a TOFD detection defect identification model, and combining the abnormal welding seam classification model and the TOFD detection defect identification model to form a final TOFD detection defect efficient identification model which can quickly and accurately realize automatic identification of TOFD detection defects. Defects are positioned, classified and quantitatively measured, the efficiency, the accuracy, the objec