Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism

Face anti-spoofing (FAS) is important to secure face recognition systems. Deep learning has obtained great success in this area, however, most existing approaches fail to consider comprehensive relation-aware local representations of live and spoof faces. To address this issue, we propose a Transfor...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on biometrics, behavior, and identity science behavior, and identity science, 2022-07, Vol.4 (3), p.439-450
Hauptverfasser: Wang, Zhuo, Wang, Qiangchang, Deng, Weihong, Guo, Guodong
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 450
container_issue 3
container_start_page 439
container_title IEEE transactions on biometrics, behavior, and identity science
container_volume 4
creator Wang, Zhuo
Wang, Qiangchang
Deng, Weihong
Guo, Guodong
description Face anti-spoofing (FAS) is important to secure face recognition systems. Deep learning has obtained great success in this area, however, most existing approaches fail to consider comprehensive relation-aware local representations of live and spoof faces. To address this issue, we propose a Transformer-based Face Anti-Spoofing (TransFAS) model to explore comprehensive facial parts for FAS. Besides the multi-head self-attention which explores relations among local patches in the same layer, we propose cross-layer relation-aware attentions (CRA) to adaptively integrate local patches from different layers. Furthermore, to effectively fuse hierarchical features, we explore the best hierarchical feature fusion (HFF) structure, which can capture the complementary information between low-level artifacts and high-level semantic features for the spoofing patterns. With these novel modules, TransFAS not only improves the generalization capability of the classical vision transformer, but also achieves SOTA performance on multiple benchmarks, demonstrating the superiority of the transformer-based model for FAS.
doi_str_mv 10.1109/TBIOM.2022.3184500
format Article
fullrecord <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_crossref_primary_10_1109_TBIOM_2022_3184500</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>9817442</ieee_id><sourcerecordid>2691875219</sourcerecordid><originalsourceid>FETCH-LOGICAL-c295t-7d1fd8edbe8eede46537e00d9b7e50e7ad535b330721154dac26a5aae63b52883</originalsourceid><addsrcrecordid>eNpNkE1Lw0AQhhdRsNT-Ab0EPKfOfmWzx1qsFloK2uJx2WQnNqXN1t0U8d-b2iLCMDOH95mBh5BbCkNKQT8sH6eL-ZABY0NOcyEBLkiPZVylmQB1-W-_JoMYNwDAQOiuemQ8sSUmo6at07e991XdfCSreOzLYJtY-bDDEJP3ul0nr7i1be2bdPRlAyZzLNe2qePuhlxVdhtxcJ59spo8Lccv6WzxPB2PZmnJtGxT5WjlcnQF5ogORSa5QgCnC4USUFknuSw4B8UolcLZkmVWWosZLyTLc94n96e7--A_Dxhbs_GH0HQvDcs0zZVkVHcpdkqVwccYsDL7UO9s-DYUzNGX-fVljr7M2VcH3Z2gGhH_AJ1TJQTjP5bWZgM</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2691875219</pqid></control><display><type>article</type><title>Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism</title><source>IEEE Electronic Library (IEL)</source><creator>Wang, Zhuo ; Wang, Qiangchang ; Deng, Weihong ; Guo, Guodong</creator><creatorcontrib>Wang, Zhuo ; Wang, Qiangchang ; Deng, Weihong ; Guo, Guodong</creatorcontrib><description>Face anti-spoofing (FAS) is important to secure face recognition systems. Deep learning has obtained great success in this area, however, most existing approaches fail to consider comprehensive relation-aware local representations of live and spoof faces. To address this issue, we propose a Transformer-based Face Anti-Spoofing (TransFAS) model to explore comprehensive facial parts for FAS. Besides the multi-head self-attention which explores relations among local patches in the same layer, we propose cross-layer relation-aware attentions (CRA) to adaptively integrate local patches from different layers. Furthermore, to effectively fuse hierarchical features, we explore the best hierarchical feature fusion (HFF) structure, which can capture the complementary information between low-level artifacts and high-level semantic features for the spoofing patterns. With these novel modules, TransFAS not only improves the generalization capability of the classical vision transformer, but also achieves SOTA performance on multiple benchmarks, demonstrating the superiority of the transformer-based model for FAS.</description><identifier>ISSN: 2637-6407</identifier><identifier>EISSN: 2637-6407</identifier><identifier>DOI: 10.1109/TBIOM.2022.3184500</identifier><language>eng</language><publisher>Piscataway: IEEE</publisher><subject>cross-layer relation-aware attentions ; Face anti-spoofing ; Face recognition ; Faces ; Feature extraction ; hierarchical feature fusion ; Machine learning ; Representation learning ; Semantics ; Spoofing ; Structural hierarchy ; Task analysis ; transformer networks ; Transformers</subject><ispartof>IEEE transactions on biometrics, behavior, and identity science, 2022-07, Vol.4 (3), p.439-450</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2022</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c295t-7d1fd8edbe8eede46537e00d9b7e50e7ad535b330721154dac26a5aae63b52883</citedby><cites>FETCH-LOGICAL-c295t-7d1fd8edbe8eede46537e00d9b7e50e7ad535b330721154dac26a5aae63b52883</cites><orcidid>0000-0002-8078-9077</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/9817442$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,776,780,792,27901,27902,54733</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/9817442$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Wang, Zhuo</creatorcontrib><creatorcontrib>Wang, Qiangchang</creatorcontrib><creatorcontrib>Deng, Weihong</creatorcontrib><creatorcontrib>Guo, Guodong</creatorcontrib><title>Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism</title><title>IEEE transactions on biometrics, behavior, and identity science</title><addtitle>TBIOM</addtitle><description>Face anti-spoofing (FAS) is important to secure face recognition systems. Deep learning has obtained great success in this area, however, most existing approaches fail to consider comprehensive relation-aware local representations of live and spoof faces. To address this issue, we propose a Transformer-based Face Anti-Spoofing (TransFAS) model to explore comprehensive facial parts for FAS. Besides the multi-head self-attention which explores relations among local patches in the same layer, we propose cross-layer relation-aware attentions (CRA) to adaptively integrate local patches from different layers. Furthermore, to effectively fuse hierarchical features, we explore the best hierarchical feature fusion (HFF) structure, which can capture the complementary information between low-level artifacts and high-level semantic features for the spoofing patterns. With these novel modules, TransFAS not only improves the generalization capability of the classical vision transformer, but also achieves SOTA performance on multiple benchmarks, demonstrating the superiority of the transformer-based model for FAS.</description><subject>cross-layer relation-aware attentions</subject><subject>Face anti-spoofing</subject><subject>Face recognition</subject><subject>Faces</subject><subject>Feature extraction</subject><subject>hierarchical feature fusion</subject><subject>Machine learning</subject><subject>Representation learning</subject><subject>Semantics</subject><subject>Spoofing</subject><subject>Structural hierarchy</subject><subject>Task analysis</subject><subject>transformer networks</subject><subject>Transformers</subject><issn>2637-6407</issn><issn>2637-6407</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkE1Lw0AQhhdRsNT-Ab0EPKfOfmWzx1qsFloK2uJx2WQnNqXN1t0U8d-b2iLCMDOH95mBh5BbCkNKQT8sH6eL-ZABY0NOcyEBLkiPZVylmQB1-W-_JoMYNwDAQOiuemQ8sSUmo6at07e991XdfCSreOzLYJtY-bDDEJP3ul0nr7i1be2bdPRlAyZzLNe2qePuhlxVdhtxcJ59spo8Lccv6WzxPB2PZmnJtGxT5WjlcnQF5ogORSa5QgCnC4USUFknuSw4B8UolcLZkmVWWosZLyTLc94n96e7--A_Dxhbs_GH0HQvDcs0zZVkVHcpdkqVwccYsDL7UO9s-DYUzNGX-fVljr7M2VcH3Z2gGhH_AJ1TJQTjP5bWZgM</recordid><startdate>20220701</startdate><enddate>20220701</enddate><creator>Wang, Zhuo</creator><creator>Wang, Qiangchang</creator><creator>Deng, Weihong</creator><creator>Guo, Guodong</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SP</scope><scope>8FD</scope><scope>L7M</scope><orcidid>https://orcid.org/0000-0002-8078-9077</orcidid></search><sort><creationdate>20220701</creationdate><title>Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism</title><author>Wang, Zhuo ; Wang, Qiangchang ; Deng, Weihong ; Guo, Guodong</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c295t-7d1fd8edbe8eede46537e00d9b7e50e7ad535b330721154dac26a5aae63b52883</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>cross-layer relation-aware attentions</topic><topic>Face anti-spoofing</topic><topic>Face recognition</topic><topic>Faces</topic><topic>Feature extraction</topic><topic>hierarchical feature fusion</topic><topic>Machine learning</topic><topic>Representation learning</topic><topic>Semantics</topic><topic>Spoofing</topic><topic>Structural hierarchy</topic><topic>Task analysis</topic><topic>transformer networks</topic><topic>Transformers</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Wang, Zhuo</creatorcontrib><creatorcontrib>Wang, Qiangchang</creatorcontrib><creatorcontrib>Deng, Weihong</creatorcontrib><creatorcontrib>Guo, Guodong</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Electronics &amp; Communications Abstracts</collection><collection>Technology Research Database</collection><collection>Advanced Technologies Database with Aerospace</collection><jtitle>IEEE transactions on biometrics, behavior, and identity science</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Wang, Zhuo</au><au>Wang, Qiangchang</au><au>Deng, Weihong</au><au>Guo, Guodong</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism</atitle><jtitle>IEEE transactions on biometrics, behavior, and identity science</jtitle><stitle>TBIOM</stitle><date>2022-07-01</date><risdate>2022</risdate><volume>4</volume><issue>3</issue><spage>439</spage><epage>450</epage><pages>439-450</pages><issn>2637-6407</issn><eissn>2637-6407</eissn><abstract>Face anti-spoofing (FAS) is important to secure face recognition systems. Deep learning has obtained great success in this area, however, most existing approaches fail to consider comprehensive relation-aware local representations of live and spoof faces. To address this issue, we propose a Transformer-based Face Anti-Spoofing (TransFAS) model to explore comprehensive facial parts for FAS. Besides the multi-head self-attention which explores relations among local patches in the same layer, we propose cross-layer relation-aware attentions (CRA) to adaptively integrate local patches from different layers. Furthermore, to effectively fuse hierarchical features, we explore the best hierarchical feature fusion (HFF) structure, which can capture the complementary information between low-level artifacts and high-level semantic features for the spoofing patterns. With these novel modules, TransFAS not only improves the generalization capability of the classical vision transformer, but also achieves SOTA performance on multiple benchmarks, demonstrating the superiority of the transformer-based model for FAS.</abstract><cop>Piscataway</cop><pub>IEEE</pub><doi>10.1109/TBIOM.2022.3184500</doi><tpages>12</tpages><orcidid>https://orcid.org/0000-0002-8078-9077</orcidid></addata></record>
fulltext fulltext_linktorsrc
identifier ISSN: 2637-6407
ispartof IEEE transactions on biometrics, behavior, and identity science, 2022-07, Vol.4 (3), p.439-450
issn 2637-6407
2637-6407
language eng
recordid cdi_crossref_primary_10_1109_TBIOM_2022_3184500
source IEEE Electronic Library (IEL)
subjects cross-layer relation-aware attentions
Face anti-spoofing
Face recognition
Faces
Feature extraction
hierarchical feature fusion
Machine learning
Representation learning
Semantics
Spoofing
Structural hierarchy
Task analysis
transformer networks
Transformers
title Face Anti-Spoofing Using Transformers With Relation-Aware Mechanism
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T16%3A56%3A45IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Face%20Anti-Spoofing%20Using%20Transformers%20With%20Relation-Aware%20Mechanism&rft.jtitle=IEEE%20transactions%20on%20biometrics,%20behavior,%20and%20identity%20science&rft.au=Wang,%20Zhuo&rft.date=2022-07-01&rft.volume=4&rft.issue=3&rft.spage=439&rft.epage=450&rft.pages=439-450&rft.issn=2637-6407&rft.eissn=2637-6407&rft_id=info:doi/10.1109/TBIOM.2022.3184500&rft_dat=%3Cproquest_RIE%3E2691875219%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2691875219&rft_id=info:pmid/&rft_ieee_id=9817442&rfr_iscdi=true