Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation
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...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SAGE Publishing
2025-01-01
|
| Series: | Journal of Engineered Fibers and Fabrics |
| Online Access: | https://doi.org/10.1177/15589250241313158 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841549966199226368 |
|---|---|
| author | Ning Dai Xiaohan Hu Kaixin Xu Xudong Hu Yanhong Yuan Bo Cao Luhong Shi |
| author_facet | Ning Dai Xiaohan Hu Kaixin Xu Xudong Hu Yanhong Yuan Bo Cao Luhong Shi |
| author_sort | Ning Dai |
| collection | DOAJ |
| description | 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. |
| format | Article |
| id | doaj-art-df4ebb3064f2410e8a6ac5472db91c31 |
| institution | Kabale University |
| issn | 1558-9250 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | SAGE Publishing |
| record_format | Article |
| series | Journal of Engineered Fibers and Fabrics |
| spelling | doaj-art-df4ebb3064f2410e8a6ac5472db91c312025-01-10T12:03:19ZengSAGE PublishingJournal of Engineered Fibers and Fabrics1558-92502025-01-012010.1177/15589250241313158Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentationNing Dai0Xiaohan Hu1Kaixin Xu2Xudong Hu3Yanhong Yuan4Bo Cao5Luhong Shi6Zhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Sci-Tech University, Hangzhou, Zhejiang, ChinaZhejiang Kangli Automation Technology Co., Ltd., Shaoxing, Zhejiang Province, ChinaIn 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.https://doi.org/10.1177/15589250241313158 |
| spellingShingle | Ning Dai Xiaohan Hu Kaixin Xu Xudong Hu Yanhong Yuan Bo Cao Luhong Shi Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation Journal of Engineered Fibers and Fabrics |
| title | Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| title_full | Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| title_fullStr | Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| title_full_unstemmed | Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| title_short | Blind super-resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| title_sort | blind super resolution network based on local fuzzy discriminative loss for fabric data augmentation |
| url | https://doi.org/10.1177/15589250241313158 |
| work_keys_str_mv | AT ningdai blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT xiaohanhu blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT kaixinxu blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT xudonghu blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT yanhongyuan blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT bocao blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation AT luhongshi blindsuperresolutionnetworkbasedonlocalfuzzydiscriminativelossforfabricdataaugmentation |