Bibliographic Details
| Title: |
Image Segmentation Method for Water Transport Feature Detection in Fabrics via Target-Located Strategy |
| Authors: |
Yucheng Wang, Huiping Wang, Hang Mao, Suwei Gao, Qiaofeng Wei, Shujing Li, Rangtong Liu, Boyang Xu |
| Source: |
IEEE Access, Vol 13, Pp 48146-48155 (2025) |
| Publisher Information: |
IEEE, 2025. |
| Publication Year: |
2025 |
| Collection: |
LCC:Electrical engineering. Electronics. Nuclear engineering |
| Subject Terms: |
Wet zone, segmentation model, detection model, SAD method, liquid water transport, Electrical engineering. Electronics. Nuclear engineering, TK1-9971 |
| More Details: |
The comfort of apparel, particularly sportswear, is affected by the materials’ liquid water transport properties. Many studies have employed machine vision techniques (taking pictures to document changes in the wet zones of fabric) to investigate the transport characteristics of liquid water in fabrics with various topologies. Machine vision techniques fix the problems of manual detection in terms of cost and efficiency. Still, several issues, such as changes in lighting during the detection process, reflections from water droplets, and wrinkles in the fabric, can lower the segmentation accuracy. To easily differ the wet zones from dry parts of fabrics, a novel segmentation and detection method (called SAD) that combines detection models like YOLOv5 with segmentation models like SAM is proposed, after the comparison with SAD to a variety of detection models, including YOLOv5, DETR, and YOLOv3, YOLOv5 presents an optimal accurate. The results show that the accuracy of calculating the area of wet zone in fabric using the SAM-SAD method reaches 94.07%, which is the same as the manual marking method. It is better than traditional models such as Otsu, Canny, and Watershed. And the time-dependent curve is closer to the actual wetting and evaporation process of fabric, the SAD method is conducive to wet zone segmentation. |
| Document Type: |
article |
| File Description: |
electronic resource |
| Language: |
English |
| ISSN: |
2169-3536 |
| Relation: |
https://ieeexplore.ieee.org/document/10921668/; https://doaj.org/toc/2169-3536 |
| DOI: |
10.1109/ACCESS.2025.3550260 |
| Access URL: |
https://doaj.org/article/15dd19a03a264c299a5790808622aed5 |
| Accession Number: |
edsdoj.15dd19a03a264c299a5790808622aed5 |
| Database: |
Directory of Open Access Journals |
