Unsupervised fabric defect detection algorithm based on vector quantization and feature distance

Fabric defect detection is an indispensable step in textile fabric production, and many deep-learning-based methods have been proposed. However, existing supervised methods are limited by the lack of annotated datasets, and unsupervised anomaly detection methods still fail to meet the requirements o...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Textile research journal
Hlavní autori: Wei, Qiyu, Zhang, Lei, Zhan, Zhu, Li, Liqing, Wang, Jun
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: 05.06.2025
ISSN:0040-5175, 1746-7748
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract Fabric defect detection is an indispensable step in textile fabric production, and many deep-learning-based methods have been proposed. However, existing supervised methods are limited by the lack of annotated datasets, and unsupervised anomaly detection methods still fail to meet the requirements of practical applications. To address these issues, a novel unsupervised dual-scale method based on feature distance, called DSFD, is proposed. First, leveraging the periodic characteristics of fabric textures, a method for obtaining feature templates of image patches and a method for calculating anomaly scores of image patches are proposed based on a vector quantized variational autoencoder (VQ-VAE). Second, to address the issue that codebook vectors cannot be effectively activated using Euclidean distance-based quantization mechanism during model training and testing, a cosine similarity-based quantization mechanism is proposed. Ablation experiments demonstrate its effectiveness in improving defect detection performance. Finally, to enhance the robustness of the model when applied to different types of fabric images, a dual-scale method is proposed. The proposed method was compared with five state-of-the-art anomaly detection methods on three open-source datasets, achieving superior defect detection performance. It demonstrated a performance improvement of 2.1% in image-level defect detection and 1.3% in pixel-level segmentation in terms of area under the curve scores.
AbstractList Fabric defect detection is an indispensable step in textile fabric production, and many deep-learning-based methods have been proposed. However, existing supervised methods are limited by the lack of annotated datasets, and unsupervised anomaly detection methods still fail to meet the requirements of practical applications. To address these issues, a novel unsupervised dual-scale method based on feature distance, called DSFD, is proposed. First, leveraging the periodic characteristics of fabric textures, a method for obtaining feature templates of image patches and a method for calculating anomaly scores of image patches are proposed based on a vector quantized variational autoencoder (VQ-VAE). Second, to address the issue that codebook vectors cannot be effectively activated using Euclidean distance-based quantization mechanism during model training and testing, a cosine similarity-based quantization mechanism is proposed. Ablation experiments demonstrate its effectiveness in improving defect detection performance. Finally, to enhance the robustness of the model when applied to different types of fabric images, a dual-scale method is proposed. The proposed method was compared with five state-of-the-art anomaly detection methods on three open-source datasets, achieving superior defect detection performance. It demonstrated a performance improvement of 2.1% in image-level defect detection and 1.3% in pixel-level segmentation in terms of area under the curve scores.
Author Li, Liqing
Wang, Jun
Zhang, Lei
Wei, Qiyu
Zhan, Zhu
Author_xml – sequence: 1
  givenname: Qiyu
  surname: Wei
  fullname: Wei, Qiyu
  organization: Key Laboratory of Textile Science & Technology of Ministry of Education, College of Textiles, Donghua University, PR China
– sequence: 2
  givenname: Lei
  surname: Zhang
  fullname: Zhang, Lei
  organization: Key Laboratory of Textile Science & Technology of Ministry of Education, College of Textiles, Donghua University, PR China
– sequence: 3
  givenname: Zhu
  surname: Zhan
  fullname: Zhan, Zhu
  organization: School of Textiles and Fashion Art Building, Shanghai University of Engineering Science, PR China
– sequence: 4
  givenname: Liqing
  surname: Li
  fullname: Li, Liqing
  organization: Key Laboratory of Textile Science & Technology of Ministry of Education, College of Textiles, Donghua University, PR China
– sequence: 5
  givenname: Jun
  orcidid: 0000-0002-5655-6070
  surname: Wang
  fullname: Wang, Jun
  organization: Key Laboratory of Textile Science & Technology of Ministry of Education, College of Textiles, Donghua University, PR China
BookMark eNplkM1OwzAQhC1UJNLCA3DLCwTsOImdI6r4kypxoeewWW_AqHWK7VSCp8dRuXH6VjOj0WqWbOFGR4xdC34jhFK3nFe8FqouayGrshHqjGVCVU2hVKUXLJv9Yg5csGUIn5xzrZXO2NvWhelA_mgDmXyA3lvMDQ2EMSEm2NHlsHsfvY0f-7yHOZekY7JGn39N4KL9gVPMpQqCOHnKjQ0RHNIlOx9gF-jqjyu2fbh_XT8Vm5fH5_XdpkDRtrFQJExftxpb7CWS7o3kQzpakCD7ErCW6SttmhKVaqkZAJIiUCNwgyDliolTL_oxBE9Dd_B2D_67E7ybJ-r-TSR_AV7bXpI
Cites_doi 10.1007/978-3-030-68799-1_35
10.1177/0040517517743688
10.1109/CVPR52729.2023.01954
10.1109/ICECE54449.2021.9674302
10.1007/s10489-021-02981-4
10.1177/0040517519862880
10.1177/15589250221101382
10.1007/978-3-030-69544-6_23
10.1177/0040517520966733
10.1007/978-0-387-73003-5_196
10.1007/s11042-022-13470-2
10.1080/00405000.2013.836784
10.1111/cote.12624
10.1177/0040517520928604
10.1117/1.600663
10.1109/CVPR46437.2021.00954
10.1016/j.eswa.2020.114066
10.1016/j.imavis.2006.07.028
10.1007/s11042-023-16340-7
10.1016/j.patrec.2004.11.016
10.1109/ICCV.2019.00179
10.14504/ajr.8.S1.18
10.1016/j.neucom.2020.11.018
10.1109/TIM.2022.3196436
10.1109/CVPR42600.2020.01438
10.1007/s11063-019-10113-w
10.1016/j.patcog.2020.107706
10.1088/1742-6596/1948/1/012160
10.3390/s18010209
10.1109/ACCESS.2023.3282993
10.1007/s11042-022-13568-7
10.1109/COMST.2014.2336610
10.1016/j.compind.2021.103551
10.1016/S0031-3203(01)00188-1
10.1177/00405175231206820
10.1108/IJCST-11-2018-0135
10.3390/app10072511
10.3390/s18041064
10.1109/CVPR.2016.90
10.1109/ACVMOT.2005.115
10.3390/app12136823
10.1007/978-3-319-99695-0_6
10.1007/s11554-020-01023-5
ContentType Journal Article
DBID AAYXX
CITATION
DOI 10.1177/00405175251342617
DatabaseName CrossRef
DatabaseTitle CrossRef
DatabaseTitleList CrossRef
DeliveryMethod fulltext_linktorsrc
Discipline Economics
Engineering
EISSN 1746-7748
ExternalDocumentID 10_1177_00405175251342617
GroupedDBID -ET
-~X
.-4
.2L
.2N
.DC
01A
0R~
123
18M
1~K
29Q
31W
31X
31Z
4.4
54M
56W
5VS
8R4
8R5
AABCJ
AABOD
AACTG
AADIR
AAHBH
AAJPV
AAPEO
AAPII
AAQXI
AATAA
AATBZ
AAYXX
ABAWP
ABCCA
ABCJG
ABFXH
ABHQH
ABIDT
ABJCF
ABJNI
ABLUO
ABPNF
ABQKF
ABQPY
ABQXT
ABUAX
ABUJY
ACCVC
ACDXX
ACFUR
ACFZE
ACGFO
ACGFS
ACGOD
ACIWK
ACLZU
ACOXC
ACROE
ACSIQ
ACUAV
ACUIR
ACXKE
ADDLC
ADEBD
ADFRT
ADNON
ADNWM
ADRRZ
ADTOS
ADVBO
AEDFJ
AEDXQ
AENEX
AEPTA
AESZF
AEUHG
AEVPJ
AEWDL
AEWHI
AFKBI
AFKRG
AFMOU
AFQAA
AFRAH
AGDVU
AGKLV
AGNWV
AGWFA
AGWNL
AHDMH
AHWHD
AJGYC
AJHME
AJUZI
AJVBE
ALMA_UNASSIGNED_HOLDINGS
AMNSR
ANDLU
ARTOV
ATCPS
AUTPY
AUVAJ
AYAKG
AYPQM
AZFZN
B8T
B8Z
B94
BBRGL
BCU
BDDNI
BDZRT
BENPR
BLC
BMVBW
BPACV
BYIEH
CITATION
CS3
DG~
DH.
DO-
DU5
DV7
DV8
E.-
EBS
FHBDP
GROUPED_SAGE_PREMIER_JOURNAL_COLLECTION
H13
HF~
HVGLF
HZ~
J8X
K.F
M0K
N9A
O9-
OFU
P.B
P2P
Q1R
Q2X
Q7P
Q83
ROL
RXW
S01
SAUOL
SCNPE
SFB
SFC
SFK
SFT
SGU
SGV
SGZ
SHB
SPJ
SPK
SPP
SPV
SSDHQ
STM
TAE
U5U
WH7
~KM
ID FETCH-LOGICAL-c199t-7e1db598c9cb3ce8bd30f3ce9a3a3b2ac53def8d62c779e6faa53d1c8ca0dca33
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001502752100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0040-5175
IngestDate Sat Nov 29 07:48:59 EST 2025
IsPeerReviewed true
IsScholarly true
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c199t-7e1db598c9cb3ce8bd30f3ce9a3a3b2ac53def8d62c779e6faa53d1c8ca0dca33
ORCID 0000-0002-5655-6070
ParticipantIDs crossref_primary_10_1177_00405175251342617
PublicationCentury 2000
PublicationDate 2025-06-05
PublicationDateYYYYMMDD 2025-06-05
PublicationDate_xml – month: 06
  year: 2025
  text: 2025-06-05
  day: 05
PublicationDecade 2020
PublicationTitle Textile research journal
PublicationYear 2025
References e_1_3_2_26_2
e_1_3_2_49_2
e_1_3_2_28_2
e_1_3_2_41_2
e_1_3_2_20_2
e_1_3_2_43_2
Zhang Y (e_1_3_2_45_2) 2023
e_1_3_2_24_2
e_1_3_2_47_2
e_1_3_2_9_2
e_1_3_2_16_2
e_1_3_2_37_2
e_1_3_2_7_2
e_1_3_2_18_2
e_1_3_2_39_2
Tao X (e_1_3_2_23_2) 2022; 71
e_1_3_2_10_2
e_1_3_2_31_2
e_1_3_2_5_2
Samariya D (e_1_3_2_22_2) 2023; 10
e_1_3_2_12_2
e_1_3_2_33_2
e_1_3_2_3_2
e_1_3_2_14_2
e_1_3_2_35_2
e_1_3_2_50_2
e_1_3_2_27_2
e_1_3_2_48_2
e_1_3_2_29_2
e_1_3_2_40_2
e_1_3_2_21_2
e_1_3_2_42_2
e_1_3_2_44_2
e_1_3_2_25_2
Van Den Oord A (e_1_3_2_46_2) 2017; 30
e_1_3_2_15_2
e_1_3_2_38_2
e_1_3_2_8_2
e_1_3_2_17_2
e_1_3_2_6_2
e_1_3_2_19_2
e_1_3_2_30_2
e_1_3_2_32_2
e_1_3_2_51_2
e_1_3_2_11_2
e_1_3_2_34_2
e_1_3_2_4_2
e_1_3_2_13_2
e_1_3_2_36_2
e_1_3_2_2_2
References_xml – ident: e_1_3_2_39_2
  doi: 10.1007/978-3-030-68799-1_35
– ident: e_1_3_2_20_2
  doi: 10.1177/0040517517743688
– ident: e_1_3_2_51_2
  doi: 10.1109/CVPR52729.2023.01954
– ident: e_1_3_2_31_2
  doi: 10.1109/ICECE54449.2021.9674302
– ident: e_1_3_2_18_2
  doi: 10.1007/s10489-021-02981-4
– ident: e_1_3_2_29_2
  doi: 10.1177/0040517519862880
– ident: e_1_3_2_28_2
  doi: 10.1177/15589250221101382
– volume: 30
  start-page: 6307
  year: 2017
  ident: e_1_3_2_46_2
  article-title: Neural discrete representation learning
  publication-title: Adv Neur Inform Process Syst
– ident: e_1_3_2_36_2
  doi: 10.1007/978-3-030-69544-6_23
– ident: e_1_3_2_40_2
  doi: 10.1177/0040517520966733
– ident: e_1_3_2_41_2
  doi: 10.1007/978-0-387-73003-5_196
– ident: e_1_3_2_15_2
  doi: 10.1007/s11042-022-13470-2
– ident: e_1_3_2_16_2
  doi: 10.1080/00405000.2013.836784
– ident: e_1_3_2_30_2
– ident: e_1_3_2_34_2
  doi: 10.1111/cote.12624
– ident: e_1_3_2_12_2
  doi: 10.1177/0040517520928604
– ident: e_1_3_2_2_2
  doi: 10.1117/1.600663
– ident: e_1_3_2_47_2
– ident: e_1_3_2_43_2
  doi: 10.1109/CVPR46437.2021.00954
– ident: e_1_3_2_49_2
  doi: 10.1016/j.eswa.2020.114066
– ident: e_1_3_2_6_2
  doi: 10.1016/j.imavis.2006.07.028
– ident: e_1_3_2_19_2
  doi: 10.1007/s11042-023-16340-7
– year: 2023
  ident: e_1_3_2_45_2
  article-title: ECF-STPM: A robust crack detection method for railway catenary components
  publication-title: IEEE Trans Instrum Meas
– ident: e_1_3_2_4_2
  doi: 10.1016/j.patrec.2004.11.016
– ident: e_1_3_2_32_2
  doi: 10.1109/ICCV.2019.00179
– ident: e_1_3_2_26_2
  doi: 10.14504/ajr.8.S1.18
– ident: e_1_3_2_42_2
  doi: 10.1016/j.neucom.2020.11.018
– volume: 71
  start-page: 1
  year: 2022
  ident: e_1_3_2_23_2
  article-title: Deep learning for unsupervised anomaly localization in industrial images: A survey
  publication-title: IEEE Trans Instrum Meas
  doi: 10.1109/TIM.2022.3196436
– ident: e_1_3_2_33_2
  doi: 10.1109/CVPR42600.2020.01438
– ident: e_1_3_2_14_2
  doi: 10.1007/s11063-019-10113-w
– ident: e_1_3_2_50_2
  doi: 10.1016/j.patcog.2020.107706
– ident: e_1_3_2_11_2
  doi: 10.1088/1742-6596/1948/1/012160
– ident: e_1_3_2_37_2
  doi: 10.3390/s18010209
– ident: e_1_3_2_21_2
  doi: 10.1109/ACCESS.2023.3282993
– volume: 10
  start-page: 829
  issue: 3
  year: 2023
  ident: e_1_3_2_22_2
  article-title: A comprehensive survey of anomaly detection algorithms
  publication-title: Ann Data Sci
– ident: e_1_3_2_17_2
  doi: 10.1007/s11042-022-13568-7
– ident: e_1_3_2_48_2
  doi: 10.1109/COMST.2014.2336610
– ident: e_1_3_2_10_2
  doi: 10.1016/j.compind.2021.103551
– ident: e_1_3_2_3_2
  doi: 10.1016/S0031-3203(01)00188-1
– ident: e_1_3_2_44_2
  doi: 10.1177/00405175231206820
– ident: e_1_3_2_7_2
  doi: 10.1108/IJCST-11-2018-0135
– ident: e_1_3_2_24_2
  doi: 10.3390/app10072511
– ident: e_1_3_2_25_2
  doi: 10.3390/s18041064
– ident: e_1_3_2_38_2
  doi: 10.1109/CVPR.2016.90
– ident: e_1_3_2_5_2
  doi: 10.1109/ACVMOT.2005.115
– ident: e_1_3_2_8_2
  doi: 10.3390/app12136823
– ident: e_1_3_2_13_2
  doi: 10.1007/s10489-021-02981-4
– ident: e_1_3_2_35_2
– ident: e_1_3_2_9_2
  doi: 10.1007/978-3-319-99695-0_6
– ident: e_1_3_2_27_2
  doi: 10.1007/s11554-020-01023-5
SSID ssj0008878
Score 2.4148974
Snippet Fabric defect detection is an indispensable step in textile fabric production, and many deep-learning-based methods have been proposed. However, existing...
SourceID crossref
SourceType Index Database
Title Unsupervised fabric defect detection algorithm based on vector quantization and feature distance
WOSCitedRecordID wos001502752100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVSPB
  databaseName: SAGE Journals HSS Package 2015
  customDbUrl:
  eissn: 1746-7748
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0008878
  issn: 0040-5175
  databaseCode: AEVPJ
  dateStart: 19990101
  isFulltext: true
  titleUrlDefault: http://journals.sagepub.com/
  providerName: SAGE Publications
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1ba9RAFB7WraA-iFbFemMefHKJJDuZzMxjkYpIKRW20rd2rm6gTdfd7NL-e89ckoYtgn3wJYTDzBDyfTnnZM5lEPpIrBRMO5JRUomsNJXKZEXKDEZX1PCCaqvCYRPs6Iifnorj0ei8q4XZXLCm4dfXYvFfoQYZgO1LZ-8Bd78oCOAeQIcrwA7XfwL-pFmtF14DrMCXdFL5VHljQ49iY1ubjga_-HW1rNv55cSbMeNDBpuwf--rLJs2FWeGyIKzofenD-W0PUWSOzsD1Q5qZZJaBs0nwycM8Z6QLPCjvlnf2aE-tPVQFqIk837YYazarn93pjXtTExpyKCiQ21b5hkt4skon21UsKyswKOP3TXvqu8QQPbz_DRwvYj_w2O3tqqLz2-ZsD6xsOi6l28v8QDtTBkVfIx29g9-Hn_vrTUoWd5lVvoZKfIdmnJtLzLwXQZOyOwZepr-HvB-RP05GtlmFz3qistXu-jJoL_kC3Q-5AKOXMCRC7jnAu65gAMXMIgiF_CQCxi4gBMXcMeFl-jk68Hsy7csHamR6UKINmO2MApegxZaEW25MiR3cCMkkURNpaYEnoKbaqoZE7ZyUoKk0FzL3GhJyCs0bq4a-xph6qjWqjJ57lQpieMMPnpSWcKcFMq4PfSpe1lni9g55eyv8Ly5z-C36PEt3d6hcbtc2_food609Wr5IQH8B5I-Zww
linkProvider SAGE Publications
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Unsupervised+fabric+defect+detection+algorithm+based+on+vector+quantization+and+feature+distance&rft.jtitle=Textile+research+journal&rft.au=Wei%2C+Qiyu&rft.au=Zhang%2C+Lei&rft.au=Zhan%2C+Zhu&rft.au=Li%2C+Liqing&rft.date=2025-06-05&rft.issn=0040-5175&rft.eissn=1746-7748&rft_id=info:doi/10.1177%2F00405175251342617&rft.externalDBID=n%2Fa&rft.externalDocID=10_1177_00405175251342617
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0040-5175&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0040-5175&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0040-5175&client=summon