Adaptively Fused Attention Module for the Fabric Defect Detection
In the fabric manufacturing industry, fabric defect detection is a practical yet challenging task, due to the problem of defects with small sizes or unremarkable appearances distributed in fabric images with high resolution. Some deep‐learning‐based solutions try to tackle the aforementioned problem...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2023-02-01
|
| Series: | Advanced Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/aisy.202200151 |
| _version_ | 1797902650592722944 |
|---|---|
| author | Jin Wang Jingru Yang Guodong Lu Cheng Zhang Zhiyong Yu Ying Yang |
| author_facet | Jin Wang Jingru Yang Guodong Lu Cheng Zhang Zhiyong Yu Ying Yang |
| author_sort | Jin Wang |
| collection | DOAJ |
| description | In the fabric manufacturing industry, fabric defect detection is a practical yet challenging task, due to the problem of defects with small sizes or unremarkable appearances distributed in fabric images with high resolution. Some deep‐learning‐based solutions try to tackle the aforementioned problem but with limited achievements. Herein, a brand new module called adaptively fused attention module (AFAM) is proposed to improve the detection performance by enabling the network to concentrate more on the small or unremarkable defects in terms of 1) enhancing the feature maps both spatial‐wise and channel‐wise, 2) enhancing the attention information flow between the channel‐attention feature map and the spatial‐attention feature map, and 3) enriching background information. Experiments on the MS‐COCO dataset show that AFAM reaches higher AP than other state‐of‐the‐art (SOTA) attention modules (e.g., squeeze‐and‐excitation network [SE], efficient channel attention [ECA], convolutional block attention module [CBAM]). Meanwhile, on the Smart Diagnosis of Cloth Flaw Dataset (CloF), the Cascade R‐CNN (Resnext101) + AFAM outperforms the SOTA in the official leaderboard of the CloF dataset. Speed‐wisely, the solution achieves a 12.19‐FPS on RTX 3090, thus making it practical to be applied in the industry of fabric defect detection. |
| first_indexed | 2024-04-10T09:21:02Z |
| format | Article |
| id | doaj.art-950cf839299b47cd90b6b0032115dbf3 |
| institution | Directory Open Access Journal |
| issn | 2640-4567 |
| language | English |
| last_indexed | 2024-04-10T09:21:02Z |
| publishDate | 2023-02-01 |
| publisher | Wiley |
| record_format | Article |
| series | Advanced Intelligent Systems |
| spelling | doaj.art-950cf839299b47cd90b6b0032115dbf32023-02-20T12:54:10ZengWileyAdvanced Intelligent Systems2640-45672023-02-0152n/an/a10.1002/aisy.202200151Adaptively Fused Attention Module for the Fabric Defect DetectionJin Wang0Jingru Yang1Guodong Lu2Cheng Zhang3Zhiyong Yu4Ying Yang5State Key Laboratory of Fluid Power & Mechatronic Systems Zhejiang University Hangzhou Zhejiang 310027 ChinaState Key Laboratory of Fluid Power & Mechatronic Systems Zhejiang University Hangzhou Zhejiang 310027 ChinaState Key Laboratory of Fluid Power & Mechatronic Systems Zhejiang University Hangzhou Zhejiang 310027 ChinaState Key Laboratory of Fluid Power & Mechatronic Systems Zhejiang University Hangzhou Zhejiang 310027 ChinaState Key Laboratory of Fluid Power & Mechatronic Systems Zhejiang University Hangzhou Zhejiang 310027 ChinaZhejiang Academy of Surveying & Mapping Hangzhou Zhejiang 311100 ChinaIn the fabric manufacturing industry, fabric defect detection is a practical yet challenging task, due to the problem of defects with small sizes or unremarkable appearances distributed in fabric images with high resolution. Some deep‐learning‐based solutions try to tackle the aforementioned problem but with limited achievements. Herein, a brand new module called adaptively fused attention module (AFAM) is proposed to improve the detection performance by enabling the network to concentrate more on the small or unremarkable defects in terms of 1) enhancing the feature maps both spatial‐wise and channel‐wise, 2) enhancing the attention information flow between the channel‐attention feature map and the spatial‐attention feature map, and 3) enriching background information. Experiments on the MS‐COCO dataset show that AFAM reaches higher AP than other state‐of‐the‐art (SOTA) attention modules (e.g., squeeze‐and‐excitation network [SE], efficient channel attention [ECA], convolutional block attention module [CBAM]). Meanwhile, on the Smart Diagnosis of Cloth Flaw Dataset (CloF), the Cascade R‐CNN (Resnext101) + AFAM outperforms the SOTA in the official leaderboard of the CloF dataset. Speed‐wisely, the solution achieves a 12.19‐FPS on RTX 3090, thus making it practical to be applied in the industry of fabric defect detection.https://doi.org/10.1002/aisy.202200151adaptively fused attention modulefabric defect detectionfeature pyramid networkssmall object detectionsmart diagnosis of cloth flaw dataset |
| spellingShingle | Jin Wang Jingru Yang Guodong Lu Cheng Zhang Zhiyong Yu Ying Yang Adaptively Fused Attention Module for the Fabric Defect Detection Advanced Intelligent Systems adaptively fused attention module fabric defect detection feature pyramid networks small object detection smart diagnosis of cloth flaw dataset |
| title | Adaptively Fused Attention Module for the Fabric Defect Detection |
| title_full | Adaptively Fused Attention Module for the Fabric Defect Detection |
| title_fullStr | Adaptively Fused Attention Module for the Fabric Defect Detection |
| title_full_unstemmed | Adaptively Fused Attention Module for the Fabric Defect Detection |
| title_short | Adaptively Fused Attention Module for the Fabric Defect Detection |
| title_sort | adaptively fused attention module for the fabric defect detection |
| topic | adaptively fused attention module fabric defect detection feature pyramid networks small object detection smart diagnosis of cloth flaw dataset |
| url | https://doi.org/10.1002/aisy.202200151 |
| work_keys_str_mv | AT jinwang adaptivelyfusedattentionmoduleforthefabricdefectdetection AT jingruyang adaptivelyfusedattentionmoduleforthefabricdefectdetection AT guodonglu adaptivelyfusedattentionmoduleforthefabricdefectdetection AT chengzhang adaptivelyfusedattentionmoduleforthefabricdefectdetection AT zhiyongyu adaptivelyfusedattentionmoduleforthefabricdefectdetection AT yingyang adaptivelyfusedattentionmoduleforthefabricdefectdetection |