A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling

In order to solve the problem of high computational resource consumption in fabric anomaly detection, we propose a lightweight network, GH-YOLOx, which integrates ghost convolutions and hierarchical GHNetV2 backbone together to capture both local and global anomaly features. At the same time, other...

Full description

Saved in:
Bibliographic Details
Main Authors: Shuqin Cui, Weihong Liu, Min Li
Format: Article
Language:English
Published: MDPI AG 2025-03-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/7/2038
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849769401324142592
author Shuqin Cui
Weihong Liu
Min Li
author_facet Shuqin Cui
Weihong Liu
Min Li
author_sort Shuqin Cui
collection DOAJ
description In order to solve the problem of high computational resource consumption in fabric anomaly detection, we propose a lightweight network, GH-YOLOx, which integrates ghost convolutions and hierarchical GHNetV2 backbone together to capture both local and global anomaly features. At the same time, other innovative components, such as GhostConv, dynamic convolutions, feature fusion modules, and a shared group convolution head, are applied to effectively handle multi-scale issues. Lamp pruning accelerates inference, while channel-wise knowledge distillation enhances the pruned model’s accuracy. Experiments on fabric datasets demonstrate that GH-YOLOx can effectively reduce the number of parameters while achieving a higher detection rate than other lightweight models. Overall, our solution offers a practical approach to real-time fabric anomaly detection on mobile and embedded devices.
format Article
id doaj-art-eb6608a75b0147c890ba99e2906f31dd
institution DOAJ
issn 1424-8220
language English
publishDate 2025-03-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-eb6608a75b0147c890ba99e2906f31dd2025-08-20T03:03:25ZengMDPI AGSensors1424-82202025-03-01257203810.3390/s25072038A Lightweight Approach to Comprehensive Fabric Anomaly Detection ModelingShuqin Cui0Weihong Liu1Min Li2School of Computer and Artificial Intelligence, Wuhan Textile University, Wuhan 430072, ChinaSchool of Computer and Artificial Intelligence, Wuhan Textile University, Wuhan 430072, ChinaSchool of Computer and Artificial Intelligence, Wuhan Textile University, Wuhan 430072, ChinaIn order to solve the problem of high computational resource consumption in fabric anomaly detection, we propose a lightweight network, GH-YOLOx, which integrates ghost convolutions and hierarchical GHNetV2 backbone together to capture both local and global anomaly features. At the same time, other innovative components, such as GhostConv, dynamic convolutions, feature fusion modules, and a shared group convolution head, are applied to effectively handle multi-scale issues. Lamp pruning accelerates inference, while channel-wise knowledge distillation enhances the pruned model’s accuracy. Experiments on fabric datasets demonstrate that GH-YOLOx can effectively reduce the number of parameters while achieving a higher detection rate than other lightweight models. Overall, our solution offers a practical approach to real-time fabric anomaly detection on mobile and embedded devices.https://www.mdpi.com/1424-8220/25/7/2038lightweight networkfabric anomaly detectionlamp pruning techniqueknowledge distillation
spellingShingle Shuqin Cui
Weihong Liu
Min Li
A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
Sensors
lightweight network
fabric anomaly detection
lamp pruning technique
knowledge distillation
title A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
title_full A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
title_fullStr A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
title_full_unstemmed A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
title_short A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling
title_sort lightweight approach to comprehensive fabric anomaly detection modeling
topic lightweight network
fabric anomaly detection
lamp pruning technique
knowledge distillation
url https://www.mdpi.com/1424-8220/25/7/2038
work_keys_str_mv AT shuqincui alightweightapproachtocomprehensivefabricanomalydetectionmodeling
AT weihongliu alightweightapproachtocomprehensivefabricanomalydetectionmodeling
AT minli alightweightapproachtocomprehensivefabricanomalydetectionmodeling
AT shuqincui lightweightapproachtocomprehensivefabricanomalydetectionmodeling
AT weihongliu lightweightapproachtocomprehensivefabricanomalydetectionmodeling
AT minli lightweightapproachtocomprehensivefabricanomalydetectionmodeling