journal article·2025-01Open access

Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation

Ning Dai, Xiaohan Hu, Kaixin Xu, Xudong Hu, Yanhong Yuan, Bo Cao, Luhong Shi

Abstract

In the field of fabric defect detection, the development of algorithms has been hindered by issues such as poor quality and limited quantity of open-source datasets. Traditional data augmentation methods offer limited improvements in model performance, while generative data augmentation methods are plagued by difficulties in training generative models, susceptibility to artifacts, and the need for re-labeling. To address these challenges, this paper proposes a blind super-resolution algorithm for fabric defect data augmentation. The model is based on Real-ESRGAN and has been optimized specifically for the resolution degradation module to better adapt to the resolution degradation process in fabric images. Subsequently, a novel loss function named Local Blur Discrimination Loss is designed to address the local blur phenomenon and suppress the generation of fabric artifacts during the super-resolution process. Finally, both subjective evaluations of super-resolution effects and objective comparisons of data augmentation performance were conducted during the experimental phase. The subjective assessments demonstrate that the proposed method outperforms the baseline model. Additionally, in terms of objective performance, augmenting the DAGM2007 dataset using the proposed model, the detection model's accuracy (P) increased by 7.4%, recall (R) increased by 1.0%, and the mean average precision (mAP) increased by 2.5%, surpassing commonly used traditional vision-based data augmentation algorithms.
Source: crossref · Quality: baseline

Research context

OpenAlex topics: Industrial Vision Systems and Defect Detection · Advanced Image Processing Techniques · Image Processing Techniques and Applications

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

Zhejiang Sci-Tech University, Hangzhou, Zhejiang, China

Zhejiang Kangli Automation Technology Co., Ltd., Shaoxing, Zhejiang Province, China

References (51)

Show source-provided references
1. 10.1038/nature1453910.1038/nature14539 ↗
2. 10.1109/ACCESS.2021.314011810.1109/ACCESS.2021.3140118 ↗
3. 10.1109/SSCI.2018.862874210.1109/SSCI.2018.8628742 ↗
4. 10.1186/s40537-019-0197-010.1186/s40537-019-0197-0 ↗
5. 10.1016/j.aei.2023.10220510.1016/j.aei.2023.102205 ↗
6. 10.1080/00207543.2021.201082710.1080/00207543.2021.2010827 ↗
7. 10.1515/epoly-2022-007110.1515/epoly-2022-0071 ↗
8. 10.1109/MSP.2023.326290610.1109/MSP.2023.3262906 ↗
9. Xu Y · Textile Res J · 2024 · 911434727
10. Goodfellow I · Adv Neural Inform Proc Sys · 2014 · 27 · 5
11. 10.1016/j.simpat.2021.10240010.1016/j.simpat.2021.102400 ↗
12. 10.1109/TIM.2022.321428510.1109/TIM.2022.3214285 ↗
13. Wang P · 2022 · 255
14. 10.1109/TPAMI.2016.257703110.1109/TPAMI.2016.2577031 ↗
15. 10.1016/j.compstruct.2023.11705210.1016/j.compstruct.2023.117052 ↗
16. Mirza M, Osindero S. Conditional generative adversarial nets. arXiv preprint arXiv 2014; 1411: 1784.
17. Li W · IEEE transactions on automation science and engineering · 2023
18. Yao L · Computer graphics international conference · 2023 · 137
19. 10.1109/CVPR52688.2022.0019710.1109/CVPR52688.2022.00197 ↗
20. 10.1109/ACCESS.2024.337117510.1109/ACCESS.2024.3371175 ↗
21. 10.1002/col.2274510.1002/col.22745 ↗
22. 10.1016/j.pmatsci.2021.10091110.1016/j.pmatsci.2021.100911 ↗
23. 10.1186/s40537-023-00876-410.1186/s40537-023-00876-4 ↗
24. 10.1109/ICCVW54120.2021.0021710.1109/ICCVW54120.2021.00217 ↗
25. Wieler M · 2007 · 6 · 11
26. 10.1109/CVPR.2018.0026210.1109/CVPR.2018.00262 ↗
27. 10.1117/12.259287210.1117/12.2592872 ↗
28. 10.1109/CVPR.2016.9010.1109/CVPR.2016.90 ↗
29. 10.1109/ACCESS.2021.306106210.1109/ACCESS.2021.3061062 ↗
30. 10.1145/363642410.1145/3636424 ↗
31. 10.1007/978-3-319-24574-4_2810.1007/978-3-319-24574-4_28 ↗
32. 10.1007/s11220-021-00370-210.1007/s11220-021-00370-2 ↗
33. Miyato T · arXiv preprint arXiv · 2018 · 1802 · 05957
34. 10.1007/978-3-319-46475-6_4310.1007/978-3-319-46475-6_43 ↗
35. 10.1016/j.iot.2023.10105410.1016/j.iot.2023.101054 ↗
36. 10.1002/mrm.2605410.1002/mrm.26054 ↗
37. 10.1007/s11042-016-3938-510.1007/s11042-016-3938-5 ↗
38. 10.1007/978-3-031-29956-8_2110.1007/978-3-031-29956-8_21 ↗
39. 10.3788/LOP202158.021001410.3788/LOP202158.0210014 ↗
40. Zheng K · J Reinfor Plast Comp · 2023 · 5 · 1924702740
41. 10.1016/j.compscitech.2023.11039510.1016/j.compscitech.2023.110395 ↗
42. 10.1109/CVPR52688.2022.0055710.1109/CVPR52688.2022.00557 ↗
43. Andreou I · J Instrument · 2023 · 1810.1088/1748-0221/18/06/P06007 ↗
44. 10.3389/fphys.2022.88096610.3389/fphys.2022.880966 ↗
45. Wu Y · Int J Comput Meth · 2023 · 24 · 2341007
46. 10.1109/CVPRW53098.2021.0005410.1109/CVPRW53098.2021.00054 ↗
47. 10.1109/TAP.2023.329691510.1109/TAP.2023.3296915 ↗
48. 10.1016/j.jcrysgro.2022.12674910.1016/j.jcrysgro.2022.126749 ↗
49. 10.1109/TPAMI.2024.335000410.1109/TPAMI.2024.3350004 ↗
50. Moreau A · 2022 · 1347
51. 10.1109/TIM.2023.328051910.1109/TIM.2023.3280519 ↗