Generative domain-adapted adversarial auto-encoder model for enhanced ultrasonic imaging applications
Source abstract
Abstract
In this study, we propose a class-conditioned Generative Adversarial Autoencoder (cGAAE) to improve the realism of simulated ultrasonic imaging techniques, in particular the Multi-modal Total Focusing Method (M-TFM), based on the availability of both simulated and experimental TFM images. In particular, this work studied the case of the inspection of a complex geometry block representative of weld-inspection problem based on ultrasonic multi-elements probe. The cGAAE is represented by a tailored learning schema, trained in a semi-supervised fashion on a labelled mixture of synthetic (class 0) and experimental (class 1) M-TFM images, obtained under different meaningful inspection set-ups parameters (i.e., the celerity of the transverse ultrasonic wave, the specimen back-wall slope and height, the flaw tilt and heigh). That is, the cGAAE schema consists in a combination of learning stages involving class-conditioned spatial-transformers and arbitrary style transfer endows the cGAAE of powerful generative features, such as quasi real-time generation of M-TFM images by sweep of the inspection parameters. We exploited the cGAAE model to improve the realism of simulated M-TFM images and enhance the accuracy of the inverse problem, aiming at estimating the inspection parameters based on experimental acquisitions.
Source: openaire · Quality: candidate unverified · Source record ↗
Research context
OpenAlex topics: Ultrasonics and Acoustic Wave Propagation · Flow Measurement and Analysis · Generative Adversarial Networks and Image Synthesis
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (56)
Show source-provided references
1. Deep learning in automated ultrasonic NDE developments, axioms and opportunities10.1016/j.ndteint.2022.102703 ↗
2. Machine learning and NDE: Past, present, and future10.1063/1.5099819 ↗
3. Real time flaw detection and characterization in tube through partial least squares and SVR: Application to eddy current testing
4. A novel machine learning model for eddy current testing with uncertainty10.1016/j.ndteint.2018.09.010 ↗
5. Convolutional neural network for ultrasonic weldment flaw classification in noisy conditions10.1016/j.ultras.2018.12.001 ↗
6. Supervised learning strategy for classification and regression tasks applied to aeronautical structural health monitoring problems10.1016/j.ultras.2021.106372 ↗
7. Damage detection with ultrasonic guided waves using machine learning and aggregated baselines
8. Machine learning for ultrasonic nondestructive examination of welding defects: A systematic review
9. Computerized ultrasonic imaging inspection: From shallow to deep learning10.3390/s18113820 ↗
10. Deep learning for ultrasonic crack characterization in NDE
11. Fully convolutional networks for structural health monitoring through multivariate time series classification10.1186/s40323-020-00174-1 ↗
12. Defect sizing in guided wave imaging structural health monitoring using convolutional neural networks10.1016/j.ndteint.2021.102480 ↗
13. Ultrasonic defect characterisation using the scattering matrix: A performance comparison study of Bayesian inversion and machine learning schemas10.1109/TUFFC.2021.3084798 ↗
14. Machine learning at the interface of structural health monitoring and non-destructive evaluation: Machine learning in SHM and NDE10.1098/rsta.2019.0581 ↗
15. A survey on adversarial domain adaptation10.1007/s11063-022-10977-5 ↗
16. Transfer learning of impact echo signal classification from laboratory to the field
17. Defect shape classification using transfer learning in deep convolutional neural network on magneto-optical nondestructive inspection10.3390/app12157613 ↗
18. Understanding the effect of transfer learning on the automatic welding defect detection
19. Unsupervised domain adaptation by backpropagation
20. Automated defect recognition for welds using simulation assisted TFM imaging with artificial intelligence10.1007/s10921-021-00761-1 ↗
21. Generative adversarial network with object detector discriminator for enhanced defect detection on ultrasonic B-scans10.1016/j.neucom.2021.06.094 ↗
22. Dual-IRT-GAN: A defect-aware deep adversarial network to perform super-resolution tasks in infrared thermographic inspection10.1016/j.compositesb.2022.110309 ↗
23. Domain adapted deep-learning for improved ultrasonic crack characterization using limited experimental data10.1109/TUFFC.2022.3151397 ↗
24. A perspective survey on deep transfer learning for fault diagnosis in industrial scenarios: Theories, applications and challenges10.1016/j.ymssp.2021.108487 ↗
25. A style-based generator architecture for generative adversarial networks
26. Donahue Jeff, Simonyan Karen. Large scale adversarial representation learning. In: Adv. neural inf. process. syst. vol. 32. 2019.
27. Brock · 2018
28. Ulyanov · 2016
29. Adversarial autoencoders
30. Towards blending physics-based numerical simulations and seismic databases using generative adversarial network10.1016/j.cma.2020.113421 ↗
31. A deep learning framework for efficient global sensitivity analysis and SHAP values calculations applied to eddy current testing problems
32. Towards a multi-fidelity deep learning framework for a fast and realistic generation of ultrasonic images in complex geometries with the multi-modal total focusing method10.1016/j.ndteint.2023.102906 ↗
33. Defect detection using ultrasonic arrays: The multi-mode total focusing method10.1016/j.ndteint.2009.10.001 ↗
34. Mirza · 2014
35. GANs trained by a two time-scale update rule converge to a local Nash equilibrium
36. ALICE: Towards understanding adversarial learning for joint distribution matching
37. Iqbal · 2018
38. Zhang · 2018
39. Makhzani · 2018
40. Yan · 2022
41. Gatys · 2015
42. Spatial transformer networks
43. Chen · 2021
44. Image-to-image translation with conditional adversarial networks
45. Miyato Takeru, Koyama Masanori. CGANs with projection discriminator. In: 6th Int. conf. on learn. representations - conf. track proc.. 2018.
46. Chen · 2016
47. Focal frequency loss for image reconstruction and synthesis
48. Zhu · 2017
49. Heusel · 2018
50. Gulrajani · 2017
51. Miyato · 2018
52. Odena · 2018
53. Visualizing data using t-SNE
54. Image quality assessment: from error visibility to structural similarity10.1109/TIP.2003.819861 ↗
55. Abadi · 2015
56. Pan · 2021