MTGGAN: generate type-controllable magnetic tile defect data to improve defect detection performance
Source abstract
Abstract
The scarcity of training samples and label information poses a major challenge for deep learning defect detection technology. Using generative models to expand sample data is considered an effective approach to address this issue. However, existing GAN models struggle to generate high-fidelity and type-controllable magnetic tile defect samples. Therefore, this paper proposes a type-controllable magnetic tile generation method (MTGGAN) that generates data with inherent label information. The model enhances fidelity and diversity in conditional generation through dual discriminator, auxiliary classifier, and auxiliary generator mechanisms. Additionally, focal loss is introduced to address the model’s class tendency issue during training. Extensive experiments demonstrate that, compared to advanced GAN methods, samples generated by MTGGAN exhibit superior quality and diversity. Furthermore, the generated data can train superior detection models. Effectively alleviating the problem of data scarcity in defect detection.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Industrial Vision Systems and Defect Detection · Image Processing and 3D Reconstruction · Advanced Neural Network Applications
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
University of Shanghai for Science and Technology
Shanghai Zhengfei Electronic Technology CO., LTD
Zhejiang University
National University of Defense Technology
Funding
National Natural Science Foundation of China · Shanghai Industrial Collaborative Innovation Project · Key Laboratory of Space Active Opto-electronics Technology of Chinese Academy of Sciences · Shanghai Science and Technology Innovation Action Plan
References (44)
Show source-provided references
1. 10.1080/10589759.2022.206385910.1080/10589759.2022.2063859 ↗
2. 10.3390/electronics1201018510.3390/electronics12010185 ↗
3. Zhang J, Wang J. Magnetic tile surface quality recognition based on multi-scale convolution neural network and within-class mixup operation. J Comput Appl. 2021;41:275–279.
4. 10.1016/j.mtcomm.2021.10310510.1016/j.mtcomm.2021.103105 ↗
5. 10.1109/TIM.2023.329547710.1109/TIM.2023.3295477 ↗
6. Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets. Adv Neural Inf Process Syst. 2014;27:2672–2680.
7. 10.1109/ICCV.2017.32410.1109/ICCV.2017.324 ↗
8. 10.1109/TRA.2002.80564610.1109/TRA.2002.805646 ↗
9. 10.1016/j.microrel.2023.11522010.1016/j.microrel.2023.115220 ↗
10. 10.1784/insi.2005.47.10.61810.1784/insi.2005.47.10.618 ↗
11. Sangdoo Y, Dongyoon H, Sanghyuk C, et al. CutMix: regularization strategy to train strong classifiers with localizable features. 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Proceedings; Seoul, Korea (South); 2019. p. 6022–6031.
12. 10.1007/s11042-024-18868-810.1007/s11042-024-18868-8 ↗
13. Zhang H, Cisse M, Dauphin Y, et al. Mixup: beyond empirical risk minimization. International Conference on Learning Representations (ICLR); Toulon, France; 2017. p. 1–15.
14. 10.3390/s2314642210.3390/s23146422 ↗
15. van Dan Oord A, Kalchbrenner N, Kavukcuoglu K. Pixel recurrent neural networks. International Conference on Machine Learning (ICML); New York City, NY, USA; 2016. p. 1747–1756.
16. Kingma DP Welling M. Auto-encoding variational Bayes. CoRR abs/1312.6114 2013.
17. Radford A, Metz L, Chintala S. Unsupervised representation learning with deep convolutional generative adversarial networks. Neural Information Processing Systems (NIPS); Barcelona, Spain; 2016.
18. Mirza M, Osindero S. Conditional generative adversarial nets. Neural Information Processing Systems (NIPS); Montreal, Canada. 2014.
19. Odena A, Olah C, Shlens J. Conditional image synthesis with auxiliary classifier gans. International Conference on Machine Learning (ICML); Sydney, Australia. PMLR; 2017. p. 2642–2651.
20. Arjovsky M, Chintala S, Bottou L. Wasserstein generative adversarial networks. International Conference on Machine Learning (ICML); Sydney, Australia. PMLR; 2017. p. 214–223.
21. Nguyen TD, Le T, Vu HT, et al. Dual discriminator generative adversarial nets. Neural Information Processing Systems (NIPS); Long Beach, California. 2017. p. 2670–2680.
22. Karras T, Aila T, Laine S, et al. Progressive growing of GANs for improved quality, stability, and variation. International Conference on Learning Representations (ICLR); Vancouver, Canada. 2017. p. 1–26.
23. Karras T, Laine S, Aila T. A style-based generator architecture for generative adversarial networks. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); Long Beach, CA, USA. 2018. p. 4396–4405.
24. 10.1109/TIM.2023.328413910.1109/TIM.2023.3284139 ↗
25. 10.1177/0040517524123747910.1177/00405175241237479 ↗
26. 10.1109/TASE.2020.296741510.1109/TASE.2020.2967415 ↗
27. 10.1007/s10845-020-01710-x10.1007/s10845-020-01710-x ↗
28. 10.1016/j.neucom.2022.05.02110.1016/j.neucom.2022.05.021 ↗
29. 10.1177/0040517524123394210.1177/00405175241233942 ↗
30. 10.1109/CVPR.2017.63210.1109/CVPR.2017.632 ↗
31. 10.1109/CVPR42600.2020.0081310.1109/CVPR42600.2020.00813 ↗
32. 10.1109/COASE.2018.856042310.1109/COASE.2018.8560423 ↗
33. Karras T, Aittala M, Laine S, et al. Alias-free generative adversarial networks. Neural Information Processing Systems (NIPS); Vancouver, Canada. 2021.
34. 10.1109/CVPR46437.2021.0161110.1109/CVPR46437.2021.01611 ↗
35. Salimans T, Goodfellow IJ, Zaremba W, et al. Improved techniques for training GANs. Neural Information Processing Systems (NIPS); Barcelona, Spain. 2016.
36. Heusel M, Ramsauer H, Unterthiner T, et al. GANs trained by a two time-scale update rule converge to a Nash equilibrium. Neural Information Processing Systems (NIPS); Long Beach, California. 2017.
37. 10.1109/CVPR.2016.30810.1109/CVPR.2016.308 ↗
38. 10.1109/BigData.2018.862252510.1109/BigData.2018.8622525 ↗
39. 10.1109/CVPR.2016.9010.1109/CVPR.2016.90 ↗
40. 10.1109/CVPR.2017.63410.1109/CVPR.2017.634 ↗
41. Tan M, Le QV. EfficientNet: rethinking Model scaling for convolutional neural networks. International Conference on Machine Learning (ICML); Los Angeles, California, USA. 2019. p. 6105–6114.
42. Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16x16 words: transformers for image recognition at scale. International Conference on Learning Representations (ICLR). 2021.
43. 10.1109/TIM.2019.291540410.1109/TIM.2019.2915404 ↗
44. 10.1109/TIM.2022.316954710.1109/TIM.2022.3169547 ↗