Assisting Weld Defect Detection Using U2Net: A Machine Learning Approach
Source abstract
Abstract
Weld defect detection is usually performed by expert human interpretation of radiography images. This process is time consuming and difficult. Deep learning solutions have been developed to assist interpretation. These solutions often use a U-Net convolutional neural network to distinguish between defect and non-defect pixels in the image. Transformers, like those used for ChatGPT, are widely adapted for computer vision tasks. This work describes a UNet, called U2Net, using transformer and convolutional U-Nets to encode global and local information simultaneously for weld defect segmentation in radiographs. Results show that U2Net provides beneficial defect detection capabilities as compared to four evaluated U-Nets, underscoring the benefit of coupling global and local information in weld defect detection.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Welding Techniques and Residual Stresses · Advanced Neural Network Applications · Non-Destructive Testing Techniques
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Los Alamos National Laboratory
University of Texas at Austin
Funding
Los Alamos National Laboratory
References (28)
Show source-provided references
1. 10.3390/app916331210.3390/app9163312 ↗
2. Esab.com: radiographic and ultrasonic testing of welds [internet] north America. https://www.esabna.com/us/en/education/blog/radiographic-and-ultrasonic-testing-of-welds.cfm (accessed 09 Dec. 2021).
3. 10.1016/j.measurement.2018.09.01110.1016/j.measurement.2018.09.011 ↗
4. 10.1016/j.ndteint.2003.12.00410.1016/j.ndteint.2003.12.004 ↗
5. 10.1088/1742-6596/1237/3/03200510.1088/1742-6596/1237/3/032005 ↗
6. 10.1520/SSMS2018003310.1520/SSMS20180033 ↗
7. R. Sizyakin, et al. Proceedings of Automated Visual Inspection and Machine Vision III, edited by J. Beyerer and F. Leon, Munich, 2019. pp. 11061E).
8. R. Tokime, X. Maldague, and L. Perron, Proceedings of Digital Imaging and Ultrasonics for NDT, New Orleans, LA, 23–25 2019.
9. 10.1109/ACCESS.2021.309348710.1109/ACCESS.2021.3093487 ↗
10. Y. Wang, F. Shi, and X. Tong, Proceedings of Intelligent Computing Methodologies, Proceedings, Part III, Nanchang, China, 2019.
11. O. Ronneberger, P. Fischer, and T. Brox, in Proceedings of International Conference on Medical image computer and computer-assisted intervention, Munich, Germany, edited by N. Navab, K. Hornegger, and W. Wells et al. 5–9 Oct. 2015.
12. A. Vaswani, et al. in Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, CA, 4–9 Dec. 2017. pp. 5998–6008.
13. N. Carion, F. Massa, G. Synnaeve, et al. in Processing of European Conference on Computer Vision, Virtual, edited by A. Vedaldi, H. Bischo, and T. Brox et al. 2020.
14. A. Dosovitskiy, et al. in Proceedings of International Conference on Learning Representations, Virtual, 2021.
15. 10.1016/j.ndteint.2021.10243510.1016/j.ndteint.2021.102435 ↗
16. 10.48550/arXiv.2105.0553710.48550/arXiv.2105.05537 ↗
17. O. Petit, et al. in Proceedings of International Workshop on Machine Learning in Medical Imaging, 2021. pp. 267–276).
18. Y. Zhang, et al. in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, 18–22 2018.
19. Z. Liu, et al. in Proceedings of the IEEE/CVF International Conference on Computer Vision, Virtual, 12–13 2021.
20. D. Mery, et al. J Nondestr Eval. 34 (5), 1–12 (2015).
21. N. Srivastava, et al. J Mach Learn Res. 15, 1929–1958 (2014).
22. 10.48550/arXiv.1710.0594110.48550/arXiv.1710.05941 ↗
23. N. Abraham and N. Khan, in Proceedings of 2019 IEEE 16th International Symposium on Biomedical Imaging, Venice, Italy; 8–11 2019. pp. 683–687).
24. 10.1016/j.compmedimag.2021.10202610.1016/j.compmedimag.2021.102026 ↗
25. 10.1186/s12880-015-0068-x10.1186/s12880-015-0068-x ↗
26. T. Sørensen in Biologiske Skrifter Vol. 5 Munksgaard in Komm. Copenhagen (1948) 1-34.
27. 10.2307/193240910.2307/1932409 ↗
28. 10.1371/journal.pone.009220910.1371/journal.pone.0092209 ↗