A deep learning semantic segmentation network with attention mechanism for concrete crack detection

Hang, J; Wu, Y; Li, Y; Lai, T; Zhang, J; Li, Y

A deep learning semantic segmentation network with attention mechanism for concrete crack detection

Hang, J Wu, Y Li, Y

Lai, T Zhang, J Li, Y

Permalink

Publisher:: SAGE Publications
Publication Type:: Journal Article
Citation:: Structural Health Monitoring: an international journal, 2022, pp. 147592172211261
Issue Date:: 2022-08-26

Closed Access

	Filename	Description	Size
	A deep learning semantic segmentation network with attention mechanism for concrete crack detection.pdf	Published version	6.07 MB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Hang, J
dc.contributor.author	Wu, Y
dc.contributor.author	Li, Y https://orcid.org/0000-0002-6720-8493
dc.contributor.author	Lai, T
dc.contributor.author	Zhang, J
dc.contributor.author	Li, Y https://orcid.org/0000-0002-6720-8493
dc.date.accessioned	2023-04-20T04:01:18Z
dc.date.available	2023-04-20T04:01:18Z
dc.date.issued	2022-08-26
dc.identifier.citation	Structural Health Monitoring: an international journal, 2022, pp. 147592172211261
dc.identifier.issn	1475-9217
dc.identifier.issn	1741-3168
dc.identifier.uri	http://hdl.handle.net/10453/170025
dc.description.abstract	In this research, an attention-based feature fusion network (AFFNet), with a backbone residual network (ResNet101) enhanced with two attention mechanism modules, is proposed for automatic pixel-level detection of concrete crack. In particular, the inclusion of attention mechanism modules, for example, the vertical and horizontal compression attention module (VH-CAM) and the efficient channel attention upsample module (ECAUM), is to enable selective concentration on the crack feature. The VH-CAM generates a feature map integrating pixel-level information in vertical and horizontal directions. The ECAUM applied on each decoder layer combines efficient channel attention (ECA) and feature fusion, which can provide rich contextual information as guidance to help low-level features recover crack localization. The proposed model is evaluated on the test dataset and the results reach 84.49% for mean intersection over union (MIoU). Comparison with other state-of-the-art models proves high efficiency and accuracy of the proposed method.
dc.language	en
dc.publisher	SAGE Publications
dc.relation.ispartof	Structural Health Monitoring: an international journal
dc.relation.isbasedon	10.1177/14759217221126170
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject	09 Engineering
dc.subject.classification	Acoustics
dc.title	A deep learning semantic segmentation network with attention mechanism for concrete crack detection
dc.type	Journal Article
utslib.for	09 Engineering
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/School of Civil and Environmental Engineering
pubs.organisational-group	/University of Technology Sydney/Strength - CBI - Centre for Built Infrastructure
utslib.copyright.status	closed_access	*
pubs.consider-herdc	true
dc.date.updated	2023-04-20T04:01:17Z
pubs.publication-status	Accepted

Abstract:

In this research, an attention-based feature fusion network (AFFNet), with a backbone residual network (ResNet101) enhanced with two attention mechanism modules, is proposed for automatic pixel-level detection of concrete crack. In particular, the inclusion of attention mechanism modules, for example, the vertical and horizontal compression attention module (VH-CAM) and the efficient channel attention upsample module (ECAUM), is to enable selective concentration on the crack feature. The VH-CAM generates a feature map integrating pixel-level information in vertical and horizontal directions. The ECAUM applied on each decoder layer combines efficient channel attention (ECA) and feature fusion, which can provide rich contextual information as guidance to help low-level features recover crack localization. The proposed model is evaluated on the test dataset and the results reach 84.49% for mean intersection over union (MIoU). Comparison with other state-of-the-art models proves high efficiency and accuracy of the proposed method.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/170025