| Abstract Scope |
Welding technology is essential in today's industrial landscape, where the quality inspection of welded products stands as a critical safeguard to ensure production safety. Efficient and accurate non-destructive testing (NDT) is of paramount importance. Conventional NDT methods, which typically involve X-ray inspections, depend heavily on manual evaluations by quality control staff. This approach is not only time-consuming but also prone to inaccuracies. The application of deep learning in computer vision presents a promising alternative for enhancing X-ray weld inspections. This study explores the integration of deep learning with X-ray NDT to automate the detection of weld defects, specifically focusing on butt-welded small-diameter pipes. It comprehensively examines data, models, and systems to enhance our understanding of deep learning approaches for identifying weld defects. |