Surface defect classification of steels with a new semi-supervised learning method

•A semi-supervised learning method named CAE-SGAN is proposed to classify surface defects of steels.•CAE-SGAN improves the performance of SGAN with limited training samples.•When training the discriminator of SGAN, the decoder network of CAE is not truncated.•CAE-SGAN is tested with sample images co...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Optics and lasers in engineering Jg. 117; S. 40 - 48
Hauptverfasser: Di, He, Ke, Xu, Peng, Zhou, Dongdong, Zhou
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Ltd 01.06.2019
Schlagworte:
ISSN:0143-8166, 1873-0302
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract •A semi-supervised learning method named CAE-SGAN is proposed to classify surface defects of steels.•CAE-SGAN improves the performance of SGAN with limited training samples.•When training the discriminator of SGAN, the decoder network of CAE is not truncated.•CAE-SGAN is tested with sample images collected from three different steel production lines.•CAE-SGAN provides a better way to apply deep learning methods to some industrial scenes. Defect inspection is extremely crucial to ensure the quality of steel surface. It affects not only the subsequent production, but also the quality of the end-products. However, due to the rare occurrence and appearance variations of defects, surface defect identification of steels has always been a challenging task. Recently, deep learning methods have shown outstanding performance in image classification, especially when there are enough training samples. Since most sample images of steel surface are unlabeled, a new semi-supervised learning method is proposed to classify surface defects of steels. The new method is named CAE-SGAN, as it is based on Convolutional Autoencoder (CAE) and semi-supervised Generative Adversarial Networks (SGAN). CAE-SGAN first trains a stacked CAE through massive unlabeled data. Considering the appearance variations of defects, the passthrough layer is used to help CAE extract fine-grained features. After CAE is trained, the encoder network of CAE is reserved as the feature extractor and fed into a softmax layer to form a new classifier. SGAN is introduced for semi-supervised learning to further improve the generalization ability of the new method. The classifier is trained with images collected from real production lines and images randomly generated by SGAN. Extensive experiments are carried out with samples captured from different steel production lines, and the results indicate that CAE-SGAN had yielded best performances compared with traditional methods. Especially for hot rolled plates, the classification rate is improved by around 16%.
AbstractList •A semi-supervised learning method named CAE-SGAN is proposed to classify surface defects of steels.•CAE-SGAN improves the performance of SGAN with limited training samples.•When training the discriminator of SGAN, the decoder network of CAE is not truncated.•CAE-SGAN is tested with sample images collected from three different steel production lines.•CAE-SGAN provides a better way to apply deep learning methods to some industrial scenes. Defect inspection is extremely crucial to ensure the quality of steel surface. It affects not only the subsequent production, but also the quality of the end-products. However, due to the rare occurrence and appearance variations of defects, surface defect identification of steels has always been a challenging task. Recently, deep learning methods have shown outstanding performance in image classification, especially when there are enough training samples. Since most sample images of steel surface are unlabeled, a new semi-supervised learning method is proposed to classify surface defects of steels. The new method is named CAE-SGAN, as it is based on Convolutional Autoencoder (CAE) and semi-supervised Generative Adversarial Networks (SGAN). CAE-SGAN first trains a stacked CAE through massive unlabeled data. Considering the appearance variations of defects, the passthrough layer is used to help CAE extract fine-grained features. After CAE is trained, the encoder network of CAE is reserved as the feature extractor and fed into a softmax layer to form a new classifier. SGAN is introduced for semi-supervised learning to further improve the generalization ability of the new method. The classifier is trained with images collected from real production lines and images randomly generated by SGAN. Extensive experiments are carried out with samples captured from different steel production lines, and the results indicate that CAE-SGAN had yielded best performances compared with traditional methods. Especially for hot rolled plates, the classification rate is improved by around 16%.
Author Di, He
Ke, Xu
Dongdong, Zhou
Peng, Zhou
Author_xml – sequence: 1
  givenname: He
  surname: Di
  fullname: Di, He
– sequence: 2
  givenname: Xu
  surname: Ke
  fullname: Ke, Xu
  email: xuke@ustb.edu.cn
– sequence: 3
  givenname: Zhou
  surname: Peng
  fullname: Peng, Zhou
– sequence: 4
  givenname: Zhou
  surname: Dongdong
  fullname: Dongdong, Zhou
BookMark eNqNkMtKAzEUhoNUsK0-g3mBGXMmc8vCRSneoCB4WYdMctKmTGdKkrb49k6tuHCj8MNZfR-cb0JGXd8hIdfAUmBQ3qzTfhtbFbBbphkDkTIYBmdkDHXFE8ZZNiJjBjlPaijLCzIJYc0GMgcYk5fXnbdKIzVoUUeqB1Nw1mkVXd_R3tIQEdtADy6uqKIdHmjAjUvCbot-7wIa2qLyneuWdINx1ZtLcm5VG_Dq-07J-_3d2_wxWTw_PM1ni0RzKGJi87IsyqYyVWYVF40xvBCVYHkjGlFWkJm6QFbouuaIQmlWKJVZgxmIWrFM8ympTl7t-xA8Wrn1bqP8hwQmj2nkWv6kkcc0ksEwGMjbX6R28evh6JVr_8HPTvwQBvcOvQzaYafROD9ElKZ3fzo-AX7Biac
CitedBy_id crossref_primary_10_1016_j_engappai_2022_105553
crossref_primary_10_3390_app9204222
crossref_primary_10_1109_ACCESS_2023_3251988
crossref_primary_10_1016_j_compeleceng_2024_109916
crossref_primary_10_1016_j_measurement_2023_112446
crossref_primary_10_1109_ACCESS_2020_3042874
crossref_primary_10_1111_mice_12632
crossref_primary_10_1016_j_rcim_2020_102083
crossref_primary_10_3390_app11073086
crossref_primary_10_1371_journal_pone_0279035
crossref_primary_10_1049_ipr2_12608
crossref_primary_10_1016_j_procir_2024_10_162
crossref_primary_10_3390_math10060963
crossref_primary_10_1049_ipr2_12983
crossref_primary_10_1109_ACCESS_2023_3326843
crossref_primary_10_3390_app10134598
crossref_primary_10_3390_machines9100221
crossref_primary_10_1007_s11003_021_00495_5
crossref_primary_10_1109_TASE_2022_3140784
crossref_primary_10_3390_s21206773
crossref_primary_10_1016_j_aei_2022_101672
crossref_primary_10_1109_TAI_2024_3372474
crossref_primary_10_3390_app11062606
crossref_primary_10_1016_j_eswa_2023_120758
crossref_primary_10_1016_j_petrol_2021_109901
crossref_primary_10_1089_3dp_2021_0191
crossref_primary_10_1371_journal_pone_0280363
crossref_primary_10_1016_j_isatra_2023_09_027
crossref_primary_10_1007_s43069_024_00337_5
crossref_primary_10_1016_j_matlet_2021_129707
crossref_primary_10_1002_srin_202200505
crossref_primary_10_1016_j_optlaseng_2023_107862
crossref_primary_10_1016_j_jmsy_2021_05_008
crossref_primary_10_1007_s10845_023_02115_2
crossref_primary_10_1016_j_eswa_2024_125413
crossref_primary_10_1109_TIM_2023_3271723
crossref_primary_10_1109_ACCESS_2021_3082690
crossref_primary_10_3390_app12126049
crossref_primary_10_1109_ACCESS_2024_3406438
crossref_primary_10_1007_s42243_023_01068_3
crossref_primary_10_1111_mice_12851
crossref_primary_10_32604_jai_2022_038875
crossref_primary_10_3390_s20040980
crossref_primary_10_1016_j_optlaseng_2022_107294
crossref_primary_10_4018_IJeC_316874
crossref_primary_10_1016_j_optlaseng_2020_106324
crossref_primary_10_1109_TIM_2021_3127648
crossref_primary_10_3390_sym14081563
crossref_primary_10_3390_s20185136
crossref_primary_10_1007_s10845_025_02604_6
crossref_primary_10_1155_2022_2549683
crossref_primary_10_1038_s41598_022_14971_8
crossref_primary_10_1007_s10845_022_02064_2
crossref_primary_10_3390_s20051459
crossref_primary_10_1038_s41598_022_15855_7
crossref_primary_10_1109_JSEN_2023_3303012
crossref_primary_10_1007_s00521_022_07990_z
crossref_primary_10_3390_s23198243
crossref_primary_10_1007_s11042_021_11716_z
crossref_primary_10_1109_ACCESS_2020_3003588
crossref_primary_10_1007_s00521_025_11007_w
crossref_primary_10_1007_s40747_021_00477_9
crossref_primary_10_1016_j_procir_2020_04_106
crossref_primary_10_1007_s10489_021_02917_y
crossref_primary_10_1145_3730576
crossref_primary_10_1007_s10845_023_02286_y
crossref_primary_10_1109_ACCESS_2020_2984539
crossref_primary_10_1371_journal_pone_0292082
crossref_primary_10_1007_s42243_023_00937_1
crossref_primary_10_1109_ACCESS_2020_2979755
crossref_primary_10_1155_2021_6637252
crossref_primary_10_1177_1063293X251363430
crossref_primary_10_1007_s10845_021_01758_3
crossref_primary_10_3390_app11167657
crossref_primary_10_1002_srin_202100554
crossref_primary_10_1109_ACCESS_2024_3380618
crossref_primary_10_1007_s11227_025_07036_w
crossref_primary_10_1109_ACCESS_2022_3193676
crossref_primary_10_1016_j_cie_2023_109766
crossref_primary_10_3390_coatings13010017
crossref_primary_10_1002_srin_202200284
crossref_primary_10_1109_TIM_2020_3030167
crossref_primary_10_1016_j_eswa_2022_117372
crossref_primary_10_1016_j_aei_2025_103138
crossref_primary_10_1109_TIM_2025_3548193
crossref_primary_10_1109_TIM_2019_2963555
crossref_primary_10_3390_electronics12112440
crossref_primary_10_1016_j_engappai_2024_109683
crossref_primary_10_1016_j_matpr_2022_08_489
crossref_primary_10_1108_EC_11_2019_0502
crossref_primary_10_1016_j_dsp_2022_103718
crossref_primary_10_1049_ipr2_12647
crossref_primary_10_1109_ACCESS_2021_3050484
crossref_primary_10_3390_met11030388
crossref_primary_10_1109_ACCESS_2024_3408718
crossref_primary_10_1007_s00170_025_16431_9
crossref_primary_10_1109_ACCESS_2023_3339780
crossref_primary_10_3390_a16020095
crossref_primary_10_3233_JIFS_230170
crossref_primary_10_1016_j_matpr_2022_06_474
crossref_primary_10_1016_j_optlaseng_2023_107674
crossref_primary_10_3233_JIFS_213031
crossref_primary_10_1109_TIM_2024_3492728
crossref_primary_10_3390_electronics11152304
crossref_primary_10_1109_TIM_2023_3249221
crossref_primary_10_1038_s41598_025_89558_0
crossref_primary_10_3390_ma17235873
crossref_primary_10_1016_j_neucom_2021_12_093
crossref_primary_10_1007_s00371_024_03289_3
crossref_primary_10_1109_TIM_2022_3204091
crossref_primary_10_1007_s10845_022_02000_4
crossref_primary_10_1109_JSEN_2021_3132460
crossref_primary_10_3390_app12168239
crossref_primary_10_3390_s22207846
crossref_primary_10_3390_electronics12173617
crossref_primary_10_1016_j_optlaseng_2019_105986
crossref_primary_10_1007_s00138_022_01286_x
crossref_primary_10_1109_ACCESS_2025_3544578
crossref_primary_10_1007_s00170_024_13407_z
crossref_primary_10_3390_electronics12234863
crossref_primary_10_1016_j_optlaseng_2019_105936
crossref_primary_10_1007_s10462_023_10475_7
crossref_primary_10_1016_j_jmapro_2024_02_006
crossref_primary_10_3390_su15043733
crossref_primary_10_1007_s00170_023_11963_4
crossref_primary_10_1016_j_optlaseng_2019_06_020
crossref_primary_10_1016_j_jii_2025_100958
crossref_primary_10_1080_03019233_2022_2078262
crossref_primary_10_3390_met10060846
crossref_primary_10_1016_j_mfglet_2022_10_001
crossref_primary_10_1007_s11633_025_1560_6
crossref_primary_10_1007_s13042_024_02433_8
crossref_primary_10_1080_0951192X_2021_1901319
crossref_primary_10_1109_ACCESS_2023_3271748
crossref_primary_10_3390_s23010544
crossref_primary_10_1016_j_autcon_2019_102989
crossref_primary_10_1007_s00170_021_06592_8
crossref_primary_10_1016_j_heliyon_2025_e42433
crossref_primary_10_3390_ma13204629
crossref_primary_10_1007_s11227_024_06662_0
crossref_primary_10_1080_24725854_2024_2425292
crossref_primary_10_3390_app132111708
crossref_primary_10_3390_coatings12111707
crossref_primary_10_3390_s22218194
crossref_primary_10_3390_coatings13122011
crossref_primary_10_1007_s13369_023_08240_7
crossref_primary_10_1109_ACCESS_2024_3349628
crossref_primary_10_1016_j_jii_2023_100463
crossref_primary_10_1088_1361_6501_ac4597
crossref_primary_10_3390_informatics11020025
crossref_primary_10_1016_j_measurement_2023_113091
crossref_primary_10_3390_app9153159
crossref_primary_10_1016_j_knosys_2024_112751
crossref_primary_10_1109_TIM_2023_3279422
crossref_primary_10_1109_TGRS_2021_3095922
Cites_doi 10.1109/TMI.2016.2528162
10.3390/met7080311
10.1109/ICCV.2015.169
10.1109/TPAMI.2012.231
10.1016/j.optlaseng.2018.01.010
10.1016/j.eswa.2010.09.012
10.1016/j.imavis.2015.01.001
10.1016/j.ndteint.2009.01.007
10.1117/1.JMI.3.3.034501
10.1109/JSTSP.2012.2212416
ContentType Journal Article
Copyright 2019 Elsevier Ltd
Copyright_xml – notice: 2019 Elsevier Ltd
DBID AAYXX
CITATION
DOI 10.1016/j.optlaseng.2019.01.011
DatabaseName CrossRef
DatabaseTitle CrossRef
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
Physics
EISSN 1873-0302
EndPage 48
ExternalDocumentID 10_1016_j_optlaseng_2019_01_011
S0143816618313393
GroupedDBID --K
--M
.~1
0R~
123
1B1
1RT
1~.
1~5
29N
4.4
457
4G.
5VS
7-5
71M
8P~
9JN
AABXZ
AACTN
AAEDT
AAEDW
AAEPC
AAIAV
AAIKJ
AAKOC
AALRI
AAOAW
AAQFI
AAQXK
AAXUO
ABFNM
ABJNI
ABMAC
ABNEU
ABXDB
ABXRA
ABYKQ
ACDAQ
ACFVG
ACGFS
ACNNM
ACRLP
ADBBV
ADEZE
ADMUD
ADTZH
AEBSH
AECPX
AEKER
AENEX
AEZYN
AFKWA
AFRZQ
AFTJW
AGHFR
AGUBO
AGYEJ
AHHHB
AHJVU
AIEXJ
AIKHN
AITUG
AIVDX
AJBFU
AJOXV
ALMA_UNASSIGNED_HOLDINGS
AMFUW
AMRAJ
ASPBG
AVWKF
AXJTR
AZFZN
BBWZM
BJAXD
BKOJK
BLXMC
CS3
DU5
EBS
EFJIC
EFLBG
EJD
EO8
EO9
EP2
EP3
F5P
FDB
FEDTE
FGOYB
FIRID
FNPLU
FYGXN
G-2
G-Q
G8K
GBLVA
HMV
HVGLF
HZ~
IHE
J1W
JJJVA
KOM
LY7
M38
M41
MAGPM
MO0
N9A
NDZJH
O-L
O9-
OAUVE
OGIMB
OZT
P-8
P-9
P2P
PC.
Q38
R2-
RIG
RNS
ROL
RPZ
SDF
SDG
SDP
SES
SET
SEW
SPC
SPCBC
SPD
SPG
SSM
SSQ
SST
SSZ
T5K
VOH
WUQ
XPP
ZMT
~02
~G-
9DU
AATTM
AAXKI
AAYWO
AAYXX
ABDPE
ABWVN
ACLOT
ACRPL
ACVFH
ADCNI
ADNMO
AEIPS
AEUPX
AFJKZ
AFPUW
AGQPQ
AIGII
AIIUN
AKBMS
AKRWK
AKYEP
ANKPU
APXCP
CITATION
EFKBS
~HD
ID FETCH-LOGICAL-c315t-f46656b7d72fa39bdd3597904b9b96712d85e05c883ee9ac05aa2fde2198a02c3
ISICitedReferencesCount 181
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000462106500005&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0143-8166
IngestDate Sat Nov 29 01:41:12 EST 2025
Tue Nov 18 21:38:16 EST 2025
Fri Feb 23 02:26:42 EST 2024
IsPeerReviewed true
IsScholarly true
Keywords Generative adversarial networks
Surface inspection
Semi-supervised learning
Convolutional autoencoder
Defect detection
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c315t-f46656b7d72fa39bdd3597904b9b96712d85e05c883ee9ac05aa2fde2198a02c3
PageCount 9
ParticipantIDs crossref_primary_10_1016_j_optlaseng_2019_01_011
crossref_citationtrail_10_1016_j_optlaseng_2019_01_011
elsevier_sciencedirect_doi_10_1016_j_optlaseng_2019_01_011
PublicationCentury 2000
PublicationDate June 2019
2019-06-00
PublicationDateYYYYMMDD 2019-06-01
PublicationDate_xml – month: 06
  year: 2019
  text: June 2019
PublicationDecade 2010
PublicationTitle Optics and lasers in engineering
PublicationYear 2019
Publisher Elsevier Ltd
Publisher_xml – name: Elsevier Ltd
References Lin M, Chen Q, Yan S, 2013. Network in network. arXiv
Masci (bib0026) 2011
Ng (bib0022) 2015
Li (bib0025) 2014
Huynh, Li, Giger (bib0023) 2016
Ren (bib0018) 2015
Paulraj, Shukry, Yaacob, Adom, Krishnan (bib0006) 2010, May
Farabet, Couprie, Najman, LeCun (bib0028) 2013; 35
Ravikumar, Ramachandran, Sugumaran (bib0001) 2011; 38
LeCun (bib0009) 1998
He (bib0016) 2014
Suvdaa, Ahn, Ko (bib0004) 2012; 6
Simonyan K, Zisserman A, 2014. Very deep convolutional networks for large-scale image recognition. arXiv
Huang, Liu, Van Der Maaten, Weinberger (bib0015) 2017; 1
Odena, A. “Semi-supervised learning with generative adversarial networks.” arXiv
Badrinarayanan, V., A. Kendall, and R. Cipolla. “Segnet: a deep convolutional encoder-decoder architecture for image segmentation.” arXiv
Goodfellow (bib0029) 2014
Landstrom, Thurley (bib0008) 2012; 6
Krizhevsky, Sutskever, Hinton (bib0010) 2012
Hoo-Chang (bib0024) 2016; 35
(2015).
.
Szegedy (bib0014) 2017
Tian, Xu (bib0002) 2017; 7
Xu, Xu, Zhou, Wang (bib0003) 2018; 105
Yun, Choi, Kim, Kim (bib0007) 2009; 42
Girshick, R. “Fast r-cnn.” arXiv
Szegedy (bib0013) 2015
Liu (bib0021) 2016
(2016).
Redmon, J., and A. Farhadi. “YOLO9000: better, faster, stronger.” arXiv preprint (2017).
Redmon (bib0019) 2016
Xu, Liu, Ai (bib0005) 2015; 35
LeCun (10.1016/j.optlaseng.2019.01.011_bib0009) 1998
Goodfellow (10.1016/j.optlaseng.2019.01.011_bib0029) 2014
Huynh (10.1016/j.optlaseng.2019.01.011_bib0023) 2016
Ren (10.1016/j.optlaseng.2019.01.011_bib0018) 2015
Ng (10.1016/j.optlaseng.2019.01.011_bib0022) 2015
Redmon (10.1016/j.optlaseng.2019.01.011_bib0019) 2016
Suvdaa (10.1016/j.optlaseng.2019.01.011_bib0004) 2012; 6
10.1016/j.optlaseng.2019.01.011_bib0027
Szegedy (10.1016/j.optlaseng.2019.01.011_bib0013) 2015
Xu (10.1016/j.optlaseng.2019.01.011_bib0003) 2018; 105
Liu (10.1016/j.optlaseng.2019.01.011_bib0021) 2016
He (10.1016/j.optlaseng.2019.01.011_bib0016) 2014
Szegedy (10.1016/j.optlaseng.2019.01.011_bib0014) 2017
10.1016/j.optlaseng.2019.01.011_bib0020
Ravikumar (10.1016/j.optlaseng.2019.01.011_bib0001) 2011; 38
Yun (10.1016/j.optlaseng.2019.01.011_bib0007) 2009; 42
Masci (10.1016/j.optlaseng.2019.01.011_bib0026) 2011
Krizhevsky (10.1016/j.optlaseng.2019.01.011_bib0010) 2012
Xu (10.1016/j.optlaseng.2019.01.011_bib0005) 2015; 35
Paulraj (10.1016/j.optlaseng.2019.01.011_bib0006) 2010
10.1016/j.optlaseng.2019.01.011_bib0017
10.1016/j.optlaseng.2019.01.011_bib0012
Tian (10.1016/j.optlaseng.2019.01.011_bib0002) 2017; 7
10.1016/j.optlaseng.2019.01.011_bib0011
Hoo-Chang (10.1016/j.optlaseng.2019.01.011_bib0024) 2016; 35
10.1016/j.optlaseng.2019.01.011_bib0030
Huang (10.1016/j.optlaseng.2019.01.011_bib0015) 2017; 1
Farabet (10.1016/j.optlaseng.2019.01.011_bib0028) 2013; 35
Landstrom (10.1016/j.optlaseng.2019.01.011_bib0008) 2012; 6
Li (10.1016/j.optlaseng.2019.01.011_bib0025) 2014
References_xml – volume: 105
  start-page: 110
  year: 2018
  end-page: 117
  ident: bib0003
  article-title: Application of RNAMlet to surface defect identification of steels
  publication-title: Opt Lasers Eng
– year: 2017
  ident: bib0014
  article-title: Inception-v4, Inception-ResNet and the impact of residual connections on learning
  publication-title: AAAI
– reference: Redmon, J., and A. Farhadi. “YOLO9000: better, faster, stronger.” arXiv preprint (2017).
– year: 2014
  ident: bib0025
  article-title: Medical image classification with convolutional neural network
  publication-title: Control Automation Robotics & Vision (ICARCV), 2014 13th International Conference on
– reference: Lin M, Chen Q, Yan S, 2013. Network in network. arXiv:
– reference: Simonyan K, Zisserman A, 2014. Very deep convolutional networks for large-scale image recognition. arXiv:
– year: 2016
  ident: bib0021
  article-title: Ssd: single shot multibox detector
  publication-title: European conference on computer vision
– volume: 7
  start-page: 311
  year: 2017
  ident: bib0002
  article-title: An algorithm for surface defect identification of steel plates based on genetic algorithm and extreme learning machine
  publication-title: Metals
– volume: 1
  start-page: 3
  year: 2017
  ident: bib0015
  article-title: Densely connected convolutional networks
  publication-title: CVPR
– start-page: 2278
  year: 1998
  end-page: 2324
  ident: bib0009
  article-title: Gradient-based learning applied to document recognition
  publication-title: Proceedings of the IEEE 86.11
– year: 2015
  ident: bib0018
  article-title: Faster r-cnn: towards real-time object detection with region proposal networks
  publication-title: Adv Neural Inf Process Syst
– reference: (2016).
– year: 2015
  ident: bib0022
  article-title: Deep learning for emotion recognition on small datasets using transfer learning
  publication-title: Proceedings of the 2015 ACM on international conference on multimodal interaction
– volume: 6
  start-page: 866
  year: 2012
  end-page: 875
  ident: bib0008
  article-title: Morphology-based crack detection for steel slabs
  publication-title: IEEE J Sel Top Signal Process
– volume: 35
  start-page: 1915
  year: 2013
  end-page: 1929
  ident: bib0028
  article-title: Learning hierarchical features for scene labeling
  publication-title: IEEE PAMI
– year: 2016
  ident: bib0023
  article-title: Digital mammographic tumor classification using transfer learning from deep convolutional neural networks
  publication-title: J Med Imaging 3.3
– start-page: 1
  year: 2010, May
  end-page: 6
  ident: bib0006
  article-title: Structural steel plate damage detection using DFT spectral energy and artificial neural network
  publication-title: Signal Processing and Its Applications (CSPA), 2010 6th International Colloquium on
– reference: (2015).
– year: 2014
  ident: bib0016
  article-title: Spatial pyramid pooling in deep convolutional networks for visual recognition
  publication-title: European conference on computer vision
– reference: Girshick, R. “Fast r-cnn.” arXiv:
– reference: Badrinarayanan, V., A. Kendall, and R. Cipolla. “Segnet: a deep convolutional encoder-decoder architecture for image segmentation.” arXiv:
– year: 2015
  ident: bib0013
  article-title: Going deeper with convolutions
  publication-title: Proceedings of the IEEE conference on computer vision and pattern recognition
– volume: 42
  start-page: 389
  year: 2009
  end-page: 397
  ident: bib0007
  article-title: Automatic detection of cracks in raw steel block using Gabor filter optimized by univariate dynamic encoding algorithm for searches (uDEAS)
  publication-title: NDT & E Int
– reference: .
– volume: 6
  start-page: 161
  year: 2012
  end-page: 165
  ident: bib0004
  article-title: Steel surface defects detection and classification using SIFT and voting strategy
  publication-title: Int J Softw Eng Appl
– start-page: 1097
  year: 2012
  end-page: 1105
  ident: bib0010
  article-title: Imagenet classification with deep convolutional neural networks
  publication-title: Advances in neural information processing systems
– year: 2014
  ident: bib0029
  article-title: Generative adversarial nets
  publication-title: Adv Neural Inf Process Sys
– volume: 35
  start-page: 1285
  year: 2016
  ident: bib0024
  article-title: Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning
  publication-title: IEEE Trans Med Imaging
– year: 2016
  ident: bib0019
  article-title: You only look once: unified, real-time object detection
  publication-title: Proceedings of the IEEE conference on computer vision and pattern recognition
– year: 2011
  ident: bib0026
  article-title: Stacked convolutional auto-encoders for hierarchical feature extraction
  publication-title: International conference on artificial neural networks
– volume: 38
  start-page: 3260
  year: 2011
  end-page: 3266
  ident: bib0001
  article-title: Machine learning approach for automated visual inspection of machine components
  publication-title: Expert Syst Appl
– volume: 35
  start-page: 23
  year: 2015
  end-page: 30
  ident: bib0005
  article-title: Application of shearlet transform to classification of surface defects for metals
  publication-title: Image Vis Comput
– reference: Odena, A. “Semi-supervised learning with generative adversarial networks.” arXiv:
– volume: 35
  start-page: 1285
  issue: 5
  year: 2016
  ident: 10.1016/j.optlaseng.2019.01.011_bib0024
  article-title: Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning
  publication-title: IEEE Trans Med Imaging
  doi: 10.1109/TMI.2016.2528162
– volume: 7
  start-page: 311
  issue: 8
  year: 2017
  ident: 10.1016/j.optlaseng.2019.01.011_bib0002
  article-title: An algorithm for surface defect identification of steel plates based on genetic algorithm and extreme learning machine
  publication-title: Metals
  doi: 10.3390/met7080311
– start-page: 2278
  year: 1998
  ident: 10.1016/j.optlaseng.2019.01.011_bib0009
  article-title: Gradient-based learning applied to document recognition
– ident: 10.1016/j.optlaseng.2019.01.011_bib0027
– ident: 10.1016/j.optlaseng.2019.01.011_bib0017
  doi: 10.1109/ICCV.2015.169
– year: 2016
  ident: 10.1016/j.optlaseng.2019.01.011_bib0021
  article-title: Ssd: single shot multibox detector
– volume: 35
  start-page: 1915
  issue: 8
  year: 2013
  ident: 10.1016/j.optlaseng.2019.01.011_bib0028
  article-title: Learning hierarchical features for scene labeling
  publication-title: IEEE PAMI
  doi: 10.1109/TPAMI.2012.231
– volume: 105
  start-page: 110
  year: 2018
  ident: 10.1016/j.optlaseng.2019.01.011_bib0003
  article-title: Application of RNAMlet to surface defect identification of steels
  publication-title: Opt Lasers Eng
  doi: 10.1016/j.optlaseng.2018.01.010
– volume: 38
  start-page: 3260
  issue: 4
  year: 2011
  ident: 10.1016/j.optlaseng.2019.01.011_bib0001
  article-title: Machine learning approach for automated visual inspection of machine components
  publication-title: Expert Syst Appl
  doi: 10.1016/j.eswa.2010.09.012
– year: 2015
  ident: 10.1016/j.optlaseng.2019.01.011_bib0013
  article-title: Going deeper with convolutions
– year: 2017
  ident: 10.1016/j.optlaseng.2019.01.011_bib0014
  article-title: Inception-v4, Inception-ResNet and the impact of residual connections on learning
– ident: 10.1016/j.optlaseng.2019.01.011_bib0020
– year: 2015
  ident: 10.1016/j.optlaseng.2019.01.011_bib0022
  article-title: Deep learning for emotion recognition on small datasets using transfer learning
– ident: 10.1016/j.optlaseng.2019.01.011_bib0012
– year: 2014
  ident: 10.1016/j.optlaseng.2019.01.011_bib0029
  article-title: Generative adversarial nets
  publication-title: Adv Neural Inf Process Sys
– ident: 10.1016/j.optlaseng.2019.01.011_bib0030
– year: 2015
  ident: 10.1016/j.optlaseng.2019.01.011_bib0018
  article-title: Faster r-cnn: towards real-time object detection with region proposal networks
  publication-title: Adv Neural Inf Process Syst
– start-page: 1097
  year: 2012
  ident: 10.1016/j.optlaseng.2019.01.011_bib0010
  article-title: Imagenet classification with deep convolutional neural networks
– year: 2016
  ident: 10.1016/j.optlaseng.2019.01.011_bib0019
  article-title: You only look once: unified, real-time object detection
– volume: 1
  start-page: 3
  year: 2017
  ident: 10.1016/j.optlaseng.2019.01.011_bib0015
  article-title: Densely connected convolutional networks
– volume: 35
  start-page: 23
  year: 2015
  ident: 10.1016/j.optlaseng.2019.01.011_bib0005
  article-title: Application of shearlet transform to classification of surface defects for metals
  publication-title: Image Vis Comput
  doi: 10.1016/j.imavis.2015.01.001
– start-page: 1
  year: 2010
  ident: 10.1016/j.optlaseng.2019.01.011_bib0006
  article-title: Structural steel plate damage detection using DFT spectral energy and artificial neural network
– volume: 42
  start-page: 389
  issue: 5
  year: 2009
  ident: 10.1016/j.optlaseng.2019.01.011_bib0007
  article-title: Automatic detection of cracks in raw steel block using Gabor filter optimized by univariate dynamic encoding algorithm for searches (uDEAS)
  publication-title: NDT & E Int
  doi: 10.1016/j.ndteint.2009.01.007
– ident: 10.1016/j.optlaseng.2019.01.011_bib0011
– year: 2016
  ident: 10.1016/j.optlaseng.2019.01.011_bib0023
  article-title: Digital mammographic tumor classification using transfer learning from deep convolutional neural networks
  publication-title: J Med Imaging 3.3
  doi: 10.1117/1.JMI.3.3.034501
– year: 2014
  ident: 10.1016/j.optlaseng.2019.01.011_bib0025
  article-title: Medical image classification with convolutional neural network
– volume: 6
  start-page: 161
  issue: 2
  year: 2012
  ident: 10.1016/j.optlaseng.2019.01.011_bib0004
  article-title: Steel surface defects detection and classification using SIFT and voting strategy
  publication-title: Int J Softw Eng Appl
– year: 2011
  ident: 10.1016/j.optlaseng.2019.01.011_bib0026
  article-title: Stacked convolutional auto-encoders for hierarchical feature extraction
– volume: 6
  start-page: 866
  issue: 7
  year: 2012
  ident: 10.1016/j.optlaseng.2019.01.011_bib0008
  article-title: Morphology-based crack detection for steel slabs
  publication-title: IEEE J Sel Top Signal Process
  doi: 10.1109/JSTSP.2012.2212416
– year: 2014
  ident: 10.1016/j.optlaseng.2019.01.011_bib0016
  article-title: Spatial pyramid pooling in deep convolutional networks for visual recognition
SSID ssj0016411
Score 2.5936644
Snippet •A semi-supervised learning method named CAE-SGAN is proposed to classify surface defects of steels.•CAE-SGAN improves the performance of SGAN with limited...
SourceID crossref
elsevier
SourceType Enrichment Source
Index Database
Publisher
StartPage 40
SubjectTerms Convolutional autoencoder
Defect detection
Generative adversarial networks
Semi-supervised learning
Surface inspection
Title Surface defect classification of steels with a new semi-supervised learning method
URI https://dx.doi.org/10.1016/j.optlaseng.2019.01.011
Volume 117
WOSCitedRecordID wos000462106500005&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: PRVESC
  databaseName: Elsevier SD Freedom Collection Journals 2021
  customDbUrl:
  eissn: 1873-0302
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0016411
  issn: 0143-8166
  databaseCode: AIEXJ
  dateStart: 19950101
  isFulltext: true
  titleUrlDefault: https://www.sciencedirect.com
  providerName: Elsevier
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV3Pb9MwFLbKBtI4TDBADBjygRsKSuLEP7hNsGlwGAiGVHGJnNgZm0oSNe20y_533oudpB2TBkJIVdRactL6ff38bL_3PUJe5UiMkS0DyTgPklzHgTJwSURik1SnieZlV2xCHB_L6VR9nkyu-lyYi5moKnl5qZr_ampoA2Nj6uxfmHu4KTTAezA6XMHscP0jw39dzktdYDpUp0tcoHuM8UCDbwh2hQnRZ7VhSfHXrf15FrTLBnmjBQ901u-XuPrSqw7sp2bQdYYbY_Yvao6MooaDY-yqYduR0PHzdDlSsSOZ7z_qoe19XZ2a-lq735HAJCi-uiMxpMqMcUlu55IFeEjpJh7HtlKwAFhmnY5dLqcnVKfl5Kdmp8n5G-m7_YfzN3WzwF9enWLEnurUWD2RrytqY0Bbd14KdAZLdMXukM1YpApIcXP_w8H043AMxZPIFbT0330tQPDGx93s3qy4LCcPyLZfa9B9h5GHZGKrHXJ_RYFyh9zrIoCL9hH54nFDHW7oOm5oXVKHG4q4oZoCbug13NAeN9Th5jH5dnhw8u4o8AU3goJF6SIoEw7ufS6MiEvNVG4Mg_WmCpNc5YqLKDYytWFaSMmsVboIU63j0liY9aQO44I9IRtVXdmnhJo8hcHjIrYpS4ywuiwjK2XOjMGDZbFLeD9OWeHV6LEoyizrww7Ps2GAMxzgLIzgFe2ScOjYOEGW27u87Q2Reb_S-YsZIOi2zs_-pfNzsjX-QV6QjcV8affI3eJicdbOX3q0_QK4S6Mw
linkProvider Elsevier
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=Surface+defect+classification+of+steels+with+a+new+semi-supervised+learning+method&rft.jtitle=Optics+and+lasers+in+engineering&rft.au=Di%2C+He&rft.au=Ke%2C+Xu&rft.au=Peng%2C+Zhou&rft.au=Dongdong%2C+Zhou&rft.date=2019-06-01&rft.pub=Elsevier+Ltd&rft.issn=0143-8166&rft.eissn=1873-0302&rft.volume=117&rft.spage=40&rft.epage=48&rft_id=info:doi/10.1016%2Fj.optlaseng.2019.01.011&rft.externalDocID=S0143816618313393
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0143-8166&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0143-8166&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0143-8166&client=summon