Experimental environment.

<div><p>To improve the accuracy and efficiency of crack segmentation in ancient wooden structures, we propose a lightweight deep neural network architecture, termed SMG-Net. The core innovation of this model lies in its multi-cooperative perception mechanism. First, the proposed Structur...

Full description

Saved in:
Bibliographic Details
Main Author: Tianke Fang (22648035) (author)
Other Authors: Zhenxing Hui (22648038) (author), Zhiying Xie (6339833) (author), Peng Yu (48605) (author), Yi Gao (112914) (author), Songdi Shi (22648041) (author), Yuanrong He (22648044) (author)
Published: 2025
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1852014704409444352
author Tianke Fang (22648035)
author2 Zhenxing Hui (22648038)
Zhiying Xie (6339833)
Peng Yu (48605)
Yi Gao (112914)
Songdi Shi (22648041)
Yuanrong He (22648044)
author2_role author
author
author
author
author
author
author_facet Tianke Fang (22648035)
Zhenxing Hui (22648038)
Zhiying Xie (6339833)
Peng Yu (48605)
Yi Gao (112914)
Songdi Shi (22648041)
Yuanrong He (22648044)
author_role author
dc.creator.none.fl_str_mv Tianke Fang (22648035)
Zhenxing Hui (22648038)
Zhiying Xie (6339833)
Peng Yu (48605)
Yi Gao (112914)
Songdi Shi (22648041)
Yuanrong He (22648044)
dc.date.none.fl_str_mv 2025-11-19T18:35:17Z
dc.identifier.none.fl_str_mv 10.1371/journal.pone.0336125.t007
dc.relation.none.fl_str_mv https://figshare.com/articles/dataset/Experimental_environment_/30659322
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Biochemistry
Ecology
Science Policy
Space Science
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
statistical significance testing
range feature dependencies
made publicly available
limited computational resources
lightweight modular architecture
heritage monitoring scenarios
global structural consistency
cooperative perception mechanism
ancient wooden structures
ancient wooden components
99 %, respectively
div >< p
grained crack segmentation
fine crack details
crack segmentation
thereby improving
source code
shallow features
scale features
results confirmed
proposed structure
promote reproducibility
particularly suitable
multiple orientations
module enhances
inference speed
future research
extensive experiments
establishes long
discriminative capability
directional pooling
core innovation
constructed dataset
complex directions
coherent recognition
blurred edges
aware cross
91 %,
dc.title.none.fl_str_mv Experimental environment.
dc.type.none.fl_str_mv Dataset
info:eu-repo/semantics/publishedVersion
dataset
description <div><p>To improve the accuracy and efficiency of crack segmentation in ancient wooden structures, we propose a lightweight deep neural network architecture, termed SMG-Net. The core innovation of this model lies in its multi-cooperative perception mechanism. First, the proposed Structure-Aware Cross-directional Pooling (SACP) establishes long-range feature dependencies in multiple orientations, addressing the challenge of coherent recognition for cracks with complex directions. Second, the Multi-path Robust Feature Extraction (MRFE) module enhances the tolerance of the model to noise and blurred edges, thereby improving the discriminative capability of shallow features. Third, the Guided Semantic–Spatial Fusion (GSSFusion) mechanism enables efficient alignment and integration of multi-scale features, ensuring both fine crack details and global structural consistency in segmentation. Extensive experiments were conducted on a self-constructed dataset of cracks in ancient wooden components and the public Masonry crack dataset. SMG-Net achieved mean Intersection-over-Union (mIoU) scores of 81.12% and 87.91%, and Pixel Accuracy (PA) of 98.91% and 98.99%, respectively, significantly outperforming mainstream approaches such as U-Net, SegFormer, and Swin-UNet, with results confirmed by statistical significance testing. Moreover, SMG-Net demonstrates superior parameter efficiency and inference speed, making it particularly suitable for heritage monitoring scenarios with limited computational resources. To promote reproducibility and future research, the source code and datasets have been made publicly available at: <a href="https://github.com/HuiZhenxing/HuiZhenxing.git" target="_blank">https://github.com/HuiZhenxing/HuiZhenxing.git</a>.</p></div>
eu_rights_str_mv openAccess
id Manara_7ee680cfa280b545fcd08afc5f2eeec9
identifier_str_mv 10.1371/journal.pone.0336125.t007
network_acronym_str Manara
network_name_str ManaraRepo
oai_identifier_str oai:figshare.com:article/30659322
publishDate 2025
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Experimental environment.Tianke Fang (22648035)Zhenxing Hui (22648038)Zhiying Xie (6339833)Peng Yu (48605)Yi Gao (112914)Songdi Shi (22648041)Yuanrong He (22648044)BiochemistryEcologyScience PolicySpace ScienceEnvironmental Sciences not elsewhere classifiedBiological Sciences not elsewhere classifiedInformation Systems not elsewhere classifiedstatistical significance testingrange feature dependenciesmade publicly availablelimited computational resourceslightweight modular architectureheritage monitoring scenariosglobal structural consistencycooperative perception mechanismancient wooden structuresancient wooden components99 %, respectivelydiv >< pgrained crack segmentationfine crack detailscrack segmentationthereby improvingsource codeshallow featuresscale featuresresults confirmedproposed structurepromote reproducibilityparticularly suitablemultiple orientationsmodule enhancesinference speedfuture researchextensive experimentsestablishes longdiscriminative capabilitydirectional poolingcore innovationconstructed datasetcomplex directionscoherent recognitionblurred edgesaware cross91 %,<div><p>To improve the accuracy and efficiency of crack segmentation in ancient wooden structures, we propose a lightweight deep neural network architecture, termed SMG-Net. The core innovation of this model lies in its multi-cooperative perception mechanism. First, the proposed Structure-Aware Cross-directional Pooling (SACP) establishes long-range feature dependencies in multiple orientations, addressing the challenge of coherent recognition for cracks with complex directions. Second, the Multi-path Robust Feature Extraction (MRFE) module enhances the tolerance of the model to noise and blurred edges, thereby improving the discriminative capability of shallow features. Third, the Guided Semantic–Spatial Fusion (GSSFusion) mechanism enables efficient alignment and integration of multi-scale features, ensuring both fine crack details and global structural consistency in segmentation. Extensive experiments were conducted on a self-constructed dataset of cracks in ancient wooden components and the public Masonry crack dataset. SMG-Net achieved mean Intersection-over-Union (mIoU) scores of 81.12% and 87.91%, and Pixel Accuracy (PA) of 98.91% and 98.99%, respectively, significantly outperforming mainstream approaches such as U-Net, SegFormer, and Swin-UNet, with results confirmed by statistical significance testing. Moreover, SMG-Net demonstrates superior parameter efficiency and inference speed, making it particularly suitable for heritage monitoring scenarios with limited computational resources. To promote reproducibility and future research, the source code and datasets have been made publicly available at: <a href="https://github.com/HuiZhenxing/HuiZhenxing.git" target="_blank">https://github.com/HuiZhenxing/HuiZhenxing.git</a>.</p></div>2025-11-19T18:35:17ZDatasetinfo:eu-repo/semantics/publishedVersiondataset10.1371/journal.pone.0336125.t007https://figshare.com/articles/dataset/Experimental_environment_/30659322CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/306593222025-11-19T18:35:17Z
spellingShingle Experimental environment.
Tianke Fang (22648035)
Biochemistry
Ecology
Science Policy
Space Science
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
statistical significance testing
range feature dependencies
made publicly available
limited computational resources
lightweight modular architecture
heritage monitoring scenarios
global structural consistency
cooperative perception mechanism
ancient wooden structures
ancient wooden components
99 %, respectively
div >< p
grained crack segmentation
fine crack details
crack segmentation
thereby improving
source code
shallow features
scale features
results confirmed
proposed structure
promote reproducibility
particularly suitable
multiple orientations
module enhances
inference speed
future research
extensive experiments
establishes long
discriminative capability
directional pooling
core innovation
constructed dataset
complex directions
coherent recognition
blurred edges
aware cross
91 %,
status_str publishedVersion
title Experimental environment.
title_full Experimental environment.
title_fullStr Experimental environment.
title_full_unstemmed Experimental environment.
title_short Experimental environment.
title_sort Experimental environment.
topic Biochemistry
Ecology
Science Policy
Space Science
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
statistical significance testing
range feature dependencies
made publicly available
limited computational resources
lightweight modular architecture
heritage monitoring scenarios
global structural consistency
cooperative perception mechanism
ancient wooden structures
ancient wooden components
99 %, respectively
div >< p
grained crack segmentation
fine crack details
crack segmentation
thereby improving
source code
shallow features
scale features
results confirmed
proposed structure
promote reproducibility
particularly suitable
multiple orientations
module enhances
inference speed
future research
extensive experiments
establishes long
discriminative capability
directional pooling
core innovation
constructed dataset
complex directions
coherent recognition
blurred edges
aware cross
91 %,