Unsupervised selective labeling for semi-supervised industrial defect detection

In industrial detection scenarios, achieving high accuracy typically relies on extensive labeled datasets, which are costly and time-consuming. This has motivated a shift towards semi-supervised learning (SSL), which leverages labeled and unlabeled data to improve learning efficiency and reduce anno...

Full description

Saved in:
Bibliographic Details
Published in:Journal of King Saud University. Computer and information sciences Vol. 36; no. 8; p. 102179
Main Authors: Jian Ge, Qin Qin, Shaojing Song, Jinhua Jiang, Zhiwei Shen
Format: Journal Article
Language:English
Published: Springer 01.10.2024
Subjects:
Industrial defect detection
Semi-supervised learning
Unsupervised spectral clustering labeling
ISSN:1319-1578
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract In industrial detection scenarios, achieving high accuracy typically relies on extensive labeled datasets, which are costly and time-consuming. This has motivated a shift towards semi-supervised learning (SSL), which leverages labeled and unlabeled data to improve learning efficiency and reduce annotation costs. This work proposes the unsupervised spectral clustering labeling (USCL) method to optimize SSL for industrial challenges like defect variability, rarity, and complex distributions. Integral to USCL, we employ the multi-task fusion self-supervised learning (MTSL) method to extract robust feature representations through multiple self-supervised tasks. Additionally, we introduce the Enhanced Spectral Clustering (ESC) method and a dynamic selecting function (DSF). ESC effectively integrates both local and global similarity matrices, improving clustering accuracy. The DSF maximally selects the most valuable instances for labeling, significantly enhancing the representativeness and diversity of the labeled data. USCL consistently improves various SSL methods compared to traditional instance selection methods. For example, it boosts Efficient Teacher by 5%, 6.6%, and 7.8% in mean Average Precision(mAP) on the Automotive Sealing Rings Defect Dataset, the Metallic Surface Defect Dataset, and the Printed Circuit Boards (PCB) Defect Dataset with 10% labeled data. Our work sets a new benchmark for SSL in industrial settings.
AbstractList In industrial detection scenarios, achieving high accuracy typically relies on extensive labeled datasets, which are costly and time-consuming. This has motivated a shift towards semi-supervised learning (SSL), which leverages labeled and unlabeled data to improve learning efficiency and reduce annotation costs. This work proposes the unsupervised spectral clustering labeling (USCL) method to optimize SSL for industrial challenges like defect variability, rarity, and complex distributions. Integral to USCL, we employ the multi-task fusion self-supervised learning (MTSL) method to extract robust feature representations through multiple self-supervised tasks. Additionally, we introduce the Enhanced Spectral Clustering (ESC) method and a dynamic selecting function (DSF). ESC effectively integrates both local and global similarity matrices, improving clustering accuracy. The DSF maximally selects the most valuable instances for labeling, significantly enhancing the representativeness and diversity of the labeled data. USCL consistently improves various SSL methods compared to traditional instance selection methods. For example, it boosts Efficient Teacher by 5%, 6.6%, and 7.8% in mean Average Precision(mAP) on the Automotive Sealing Rings Defect Dataset, the Metallic Surface Defect Dataset, and the Printed Circuit Boards (PCB) Defect Dataset with 10% labeled data. Our work sets a new benchmark for SSL in industrial settings.
Author Shaojing Song
Zhiwei Shen
Jian Ge
Jinhua Jiang
Qin Qin
Author_xml – sequence: 1
  fullname: Jian Ge
  organization: School of Intelligent Manufacturing and Control Engineering, Shanghai Polytechnic University, Shanghai, 201209, China
– sequence: 2
  fullname: Qin Qin
  organization: School of Intelligent Manufacturing and Control Engineering, Shanghai Polytechnic University, Shanghai, 201209, China; Correspondence to: Shanghai Polytechnic University, 2360 Jinhai Road, Pudong New District, Shanghai, China
– sequence: 3
  fullname: Shaojing Song
  organization: School of Intelligent Manufacturing and Control Engineering, Shanghai Polytechnic University, Shanghai, 201209, China
– sequence: 4
  fullname: Jinhua Jiang
  organization: School of Intelligent Manufacturing and Control Engineering, Shanghai Polytechnic University, Shanghai, 201209, China
– sequence: 5
  fullname: Zhiwei Shen
  organization: School of Electrical Engineering and Telecommunications, UNSW Sydney, NSW, 2052, Australia
BookMark eNpNkLlqw0AURadwII7jP0ihH5AzT7OXwWQxGNzEtZjljRlFloxGMuTvoyxFqguHe25x78ii6zsk5AHoBijIx2bTfOTJp01FKz6jCpRZkCUwMCUIpW_JOueGUgpKCs7kkhyOXZ4uOFxTxlBkbNGP6YpFax22qTsVsR9mfE7lv1rqwpTHIdm2CBhnY47xW-y7e3ITbZtx_Zcrcnx5ft--lfvD6277tC89k2osQ-U4MiYEAguBWS-VRdAyaA6Ro4kyonZecHTcRGccWCek5p5xUEF4tiK7393Q26a-DOlsh8-6t6n-Af1wqu0wJt9iTT238xMU0Wou0etoUHvlDFOVARvYFypBYqk
ContentType Journal Article
DBID DOA
DOI 10.1016/j.jksuci.2024.102179
DatabaseName Open Access Journals
DatabaseTitleList
Database_xml – sequence: 1
  dbid: DOA
  name: DOAJ Directory of Open Access Journals
  url: https://www.doaj.org/
  sourceTypes: Open Website
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
ExternalDocumentID oai_doaj_org_article_0c4a0210eea846ec8f9e8c7b937291ad
GroupedDBID --K
0R~
4.4
457
5VS
AAEDT
AAEDW
AAIKJ
AAJSJ
AALRI
AASML
AAXUO
AAYWO
ABMAC
ACGFS
ADBBV
ADEZE
ADVLN
AEXQZ
AFGXO
AFJKZ
AFTJW
AGHFR
AITUG
ALMA_UNASSIGNED_HOLDINGS
AMRAJ
APXCP
BCNDV
C6C
EBS
FDB
GROUPED_DOAJ
IXB
KQ8
M41
O-L
O9-
OK1
ROL
SES
SSZ
XH2
ID FETCH-LOGICAL-c367t-d2b4e3355e13dd3ac67ae186d841f4e9f6fe8bc54eb49fb9b1ab5684c3417d5c3
IEDL.DBID DOA
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001317397400001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1319-1578
IngestDate Tue Oct 07 09:27:31 EDT 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 8
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c367t-d2b4e3355e13dd3ac67ae186d841f4e9f6fe8bc54eb49fb9b1ab5684c3417d5c3
OpenAccessLink https://doaj.org/article/0c4a0210eea846ec8f9e8c7b937291ad
ParticipantIDs doaj_primary_oai_doaj_org_article_0c4a0210eea846ec8f9e8c7b937291ad
PublicationCentury 2000
PublicationDate 2024-10-01
PublicationDateYYYYMMDD 2024-10-01
PublicationDate_xml – month: 10
  year: 2024
  text: 2024-10-01
  day: 01
PublicationDecade 2020
PublicationTitle Journal of King Saud University. Computer and information sciences
PublicationYear 2024
Publisher Springer
Publisher_xml – name: Springer
SSID ssj0001765436
Score 2.2835495
Snippet In industrial detection scenarios, achieving high accuracy typically relies on extensive labeled datasets, which are costly and time-consuming. This has...
SourceID doaj
SourceType Open Website
StartPage 102179
SubjectTerms Industrial defect detection
Semi-supervised learning
Unsupervised spectral clustering labeling
Title Unsupervised selective labeling for semi-supervised industrial defect detection
URI https://doaj.org/article/0c4a0210eea846ec8f9e8c7b937291ad
Volume 36
WOSCitedRecordID wos001317397400001&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
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV07T8MwELZQxcDCG_FWBtaIOHFiewRExYAKA5W6Rfb5IgVEqJqU39-zk0I3FqZIVhRF_nyv5LvvGLsxGgpwAuMkt1UsjEligzmVKjl3Oi2gUi6o6z_LyUTNZvp1Y9SX54T18sD9xt0mIIyvSxANhUoEVWlUIK32_5u4cd77JlJvFFPh64r0PZOhtch36XA6l-u-uUDuev9ol1BTeZgKL17AA5PrV7M_BJfxPtsdssLorn-bA7aFzSHbW09ciAYDPGIv06Zdzr15t-iiNsywIXcVEZShrzyiFJSWP-t447b6ZzpH5NCzN-jSBQZWc8ym48e3h6d4GIkQQ1bILnapFZhRjoA8cy4zUEiDXBVOCV4J1FVRobKQC7RCV1ZbbmxeKAEUrKTLITtho-arwVMWZY47i9YKkLkAoYwht-coISKj5yYVZ-zeb0g571UvSq9DHRYInXJAp_wLnfP_eMgF2_FA9RS6SzbqFku8Ytvw3dXt4joAvwLESLUo
linkProvider Directory of Open Access Journals
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+selective+labeling+for+semi-supervised+industrial+defect+detection&rft.jtitle=Journal+of+King+Saud+University.+Computer+and+information+sciences&rft.au=Jian+Ge&rft.au=Qin+Qin&rft.au=Shaojing+Song&rft.au=Jinhua+Jiang&rft.date=2024-10-01&rft.pub=Springer&rft.issn=1319-1578&rft.volume=36&rft.issue=8&rft.spage=102179&rft_id=info:doi/10.1016%2Fj.jksuci.2024.102179&rft.externalDBID=DOA&rft.externalDocID=oai_doaj_org_article_0c4a0210eea846ec8f9e8c7b937291ad
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1319-1578&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1319-1578&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1319-1578&client=summon