A high precision YOLO model for surface defect detection based on PyConv and CISBA
Abstract Defect detection is vital for product quality in industrial production, yet current surface defect detection technologies struggle with diverse defect types and complex backgrounds. The challenge intensifies with multi-scale small targets, leading to significantly reduced detection performa...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-05-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-025-91930-z |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849312111391408128 |
|---|---|
| author | Shufen Ruan Chenmei Zhan Bo Liu Quan Wan Kunfang Song |
| author_facet | Shufen Ruan Chenmei Zhan Bo Liu Quan Wan Kunfang Song |
| author_sort | Shufen Ruan |
| collection | DOAJ |
| description | Abstract Defect detection is vital for product quality in industrial production, yet current surface defect detection technologies struggle with diverse defect types and complex backgrounds. The challenge intensifies with multi-scale small targets, leading to significantly reduced detection performance. Therefore, this paper proposes the EPSC-YOLO algorithm to improve the efficiency and accuracy of defect detection. The algorithm first introduces multi-scale attention modules and uses two newly designed pyramid convolutions in the backbone network to better identify multi-scale defects; Secondly, Soft-NMS is introduced to replace traditional NMS, which can reduce information loss and improve multi-target detection accuracy by smoothing and suppressing the scores of overlapping boxes. In addition, a new convolutional attention module, CISBA, is designed to enhance the detection capability of small targets in complex backgrounds. In the end, we validate the effectiveness of EPSC-YOLO on NEU-DET and GC10-DET datasets. The experimental results show that, compared to YOLOv9c, $$mAP^{val}_{50}$$ increases by 2% and 2.4%, and $$mAP^{val}_{50:95}$$ increases by 5.1% and 2.4%, respectively. Meanwhile, EPSC-YOLO demonstrates superior accuracy and significant advantages in real-time detection of surface defects on products compared to algorithms such as YOLOv10 and MSFT-YOLO. |
| format | Article |
| id | doaj-art-5f9c1bf4bd534cb5b31d92aeb5ae5684 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-5f9c1bf4bd534cb5b31d92aeb5ae56842025-08-20T03:53:12ZengNature PortfolioScientific Reports2045-23222025-05-0115111910.1038/s41598-025-91930-zA high precision YOLO model for surface defect detection based on PyConv and CISBAShufen Ruan0Chenmei Zhan1Bo Liu2Quan Wan3Kunfang Song4The School of Mathematical and Physical Sciences, Wuhan Textile UniversityThe School of Mathematical and Physical Sciences, Wuhan Textile UniversityThe School of Mathematical and Physical Sciences, Wuhan Textile UniversityThe School of Mathematical and Physical Sciences, Wuhan Textile UniversityThe School of Computer Science and Artificial Intelligence, Wuhan Textile UniversityAbstract Defect detection is vital for product quality in industrial production, yet current surface defect detection technologies struggle with diverse defect types and complex backgrounds. The challenge intensifies with multi-scale small targets, leading to significantly reduced detection performance. Therefore, this paper proposes the EPSC-YOLO algorithm to improve the efficiency and accuracy of defect detection. The algorithm first introduces multi-scale attention modules and uses two newly designed pyramid convolutions in the backbone network to better identify multi-scale defects; Secondly, Soft-NMS is introduced to replace traditional NMS, which can reduce information loss and improve multi-target detection accuracy by smoothing and suppressing the scores of overlapping boxes. In addition, a new convolutional attention module, CISBA, is designed to enhance the detection capability of small targets in complex backgrounds. In the end, we validate the effectiveness of EPSC-YOLO on NEU-DET and GC10-DET datasets. The experimental results show that, compared to YOLOv9c, $$mAP^{val}_{50}$$ increases by 2% and 2.4%, and $$mAP^{val}_{50:95}$$ increases by 5.1% and 2.4%, respectively. Meanwhile, EPSC-YOLO demonstrates superior accuracy and significant advantages in real-time detection of surface defects on products compared to algorithms such as YOLOv10 and MSFT-YOLO.https://doi.org/10.1038/s41598-025-91930-z |
| spellingShingle | Shufen Ruan Chenmei Zhan Bo Liu Quan Wan Kunfang Song A high precision YOLO model for surface defect detection based on PyConv and CISBA Scientific Reports |
| title | A high precision YOLO model for surface defect detection based on PyConv and CISBA |
| title_full | A high precision YOLO model for surface defect detection based on PyConv and CISBA |
| title_fullStr | A high precision YOLO model for surface defect detection based on PyConv and CISBA |
| title_full_unstemmed | A high precision YOLO model for surface defect detection based on PyConv and CISBA |
| title_short | A high precision YOLO model for surface defect detection based on PyConv and CISBA |
| title_sort | high precision yolo model for surface defect detection based on pyconv and cisba |
| url | https://doi.org/10.1038/s41598-025-91930-z |
| work_keys_str_mv | AT shufenruan ahighprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT chenmeizhan ahighprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT boliu ahighprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT quanwan ahighprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT kunfangsong ahighprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT shufenruan highprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT chenmeizhan highprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT boliu highprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT quanwan highprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba AT kunfangsong highprecisionyolomodelforsurfacedefectdetectionbasedonpyconvandcisba |