DeepFlow: Large Displacement Optical Flow with Deep Matching

Optical flow computation is a key component in many computer vision systems designed for tasks such as action detection or activity recognition. However, despite several major advances over the last decade, handling large displacement in optical flow remains an open problem. Inspired by the large di...

Full description

Saved in:
Bibliographic Details
Published in2013 IEEE International Conference on Computer Vision pp. 1385 - 1392
Main Authors Weinzaepfel, Philippe, Revaud, Jerome, Harchaoui, Zaid, Schmid, Cordelia
Format Conference Proceeding Journal Article
LanguageEnglish
Published IEEE 01.12.2013
Subjects
Online AccessGet full text
ISSN1550-5499
DOI10.1109/ICCV.2013.175

Cover

Abstract Optical flow computation is a key component in many computer vision systems designed for tasks such as action detection or activity recognition. However, despite several major advances over the last decade, handling large displacement in optical flow remains an open problem. Inspired by the large displacement optical flow of Brox and Malik, our approach, termed Deep Flow, blends a matching algorithm with a variational approach for optical flow. We propose a descriptor matching algorithm, tailored to the optical flow problem, that allows to boost performance on fast motions. The matching algorithm builds upon a multi-stage architecture with 6 layers, interleaving convolutions and max-pooling, a construction akin to deep convolutional nets. Using dense sampling, it allows to efficiently retrieve quasi-dense correspondences, and enjoys a built-in smoothing effect on descriptors matches, a valuable asset for integration into an energy minimization framework for optical flow estimation. Deep Flow efficiently handles large displacements occurring in realistic videos, and shows competitive performance on optical flow benchmarks. Furthermore, it sets a new state-of-the-art on the MPI-Sintel dataset.
AbstractList Optical flow computation is a key component in many computer vision systems designed for tasks such as action detection or activity recognition. However, despite several major advances over the last decade, handling large displacement in optical flow remains an open problem. Inspired by the large displacement optical flow of Brox and Malik, our approach, termed Deep Flow, blends a matching algorithm with a variational approach for optical flow. We propose a descriptor matching algorithm, tailored to the optical flow problem, that allows to boost performance on fast motions. The matching algorithm builds upon a multi-stage architecture with 6 layers, interleaving convolutions and max-pooling, a construction akin to deep convolutional nets. Using dense sampling, it allows to efficiently retrieve quasi-dense correspondences, and enjoys a built-in smoothing effect on descriptors matches, a valuable asset for integration into an energy minimization framework for optical flow estimation. Deep Flow efficiently handles large displacements occurring in realistic videos, and shows competitive performance on optical flow benchmarks. Furthermore, it sets a new state-of-the-art on the MPI-Sintel dataset.
Author Harchaoui, Zaid
Weinzaepfel, Philippe
Revaud, Jerome
Schmid, Cordelia
Author_xml – sequence: 1
  givenname: Philippe
  surname: Weinzaepfel
  fullname: Weinzaepfel, Philippe
  email: philippe.weinzaepfel@inria.fr
  organization: INRIA, UJK, Grenoble, France
– sequence: 2
  givenname: Jerome
  surname: Revaud
  fullname: Revaud, Jerome
  email: jerome.revaud@inria.fr
  organization: INRIA, UJK, Grenoble, France
– sequence: 3
  givenname: Zaid
  surname: Harchaoui
  fullname: Harchaoui, Zaid
  email: zaid.harchaoui@inria.fr
  organization: INRIA, UJK, Grenoble, France
– sequence: 4
  givenname: Cordelia
  surname: Schmid
  fullname: Schmid, Cordelia
  email: cordelia.schmid@inria.fr
  organization: INRIA, UJK, Grenoble, France
BookMark eNo9jz1PwzAURY1UJNrCyMSSkSXBH4ljIxaUUqhU1AVYLdd5bo3cJMRpq_57EhUxPV29c690JmhU1RUgdEtwQgiWD4ui-EooJiwheXaBJiTNpaQixXSExiTLcJylUl6hSQjfGLP-xcfoaQbQzH19fIyWut1ANHOh8drADqouWjWdM9pHAxAdXbeNBjx6153ZumpzjS6t9gFu_u4Ufc5fPoq3eLl6XRTPy3hLBelijinDtsxFarU1tgRuU4MlK8Gm2hiZSbvm-ZpJIjLBxDo3lguJJaam7INkU5Scd_dVo09H7b1qWrfT7UkRrAZ55Yw5qEFe9fJ94f5caNr6Zw-hUzsXDHivK6j3QRHOpRCMCtajd2fUAcD_LM8zQgVlvwQKZUc
CODEN IEEPAD
ContentType Conference Proceeding
Journal Article
DBID 6IE
6IH
CBEJK
RIE
RIO
7SC
8FD
JQ2
L7M
L~C
L~D
ADTOC
UNPAY
DOI 10.1109/ICCV.2013.175
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan (POP) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP) 1998-present
Computer and Information Systems Abstracts
Technology Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
Unpaywall for CDI: Periodical Content
Unpaywall
DatabaseTitle Computer and Information Systems Abstracts
Technology Research Database
Computer and Information Systems Abstracts – Academic
Advanced Technologies Database with Aerospace
ProQuest Computer Science Collection
Computer and Information Systems Abstracts Professional
DatabaseTitleList
Computer and Information Systems Abstracts
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://proxy.k.utb.cz/login?url=https://ieeexplore.ieee.org/
  sourceTypes: Publisher
– sequence: 2
  dbid: UNPAY
  name: Unpaywall
  url: https://proxy.k.utb.cz/login?url=https://unpaywall.org/
  sourceTypes: Open Access Repository
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
EISBN 1479928402
9781479928408
EndPage 1392
ExternalDocumentID oai:HAL:hal-00873592v1
6751282
Genre orig-research
GroupedDBID 29O
6IE
6IF
6IH
6IK
6IL
6IM
6IN
AAJGR
AAWTH
ACGFS
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IPLJI
M43
OCL
RIE
RIL
RIO
RNS
7SC
8FD
JQ2
L7M
L~C
L~D
ADTOC
UNPAY
ID FETCH-LOGICAL-h281t-60230fd784fafcfde6f4c093def4acc959fb67b39185838b7cf6890902cdb7c93
IEDL.DBID UNPAY
ISSN 1550-5499
IngestDate Sun Sep 07 10:48:39 EDT 2025
Fri Jul 11 12:24:36 EDT 2025
Wed Aug 27 04:21:35 EDT 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed false
IsScholarly true
Language English
License other-oa
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-h281t-60230fd784fafcfde6f4c093def4acc959fb67b39185838b7cf6890902cdb7c93
Notes ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Conference-1
ObjectType-Feature-3
content type line 23
SourceType-Conference Papers & Proceedings-2
OpenAccessLink https://proxy.k.utb.cz/login?url=https://inria.hal.science/hal-00873592
PQID 1669883283
PQPubID 23500
PageCount 8
ParticipantIDs unpaywall_primary_10_1109_iccv_2013_175
ieee_primary_6751282
proquest_miscellaneous_1669883283
PublicationCentury 2000
PublicationDate 20131201
PublicationDateYYYYMMDD 2013-12-01
PublicationDate_xml – month: 12
  year: 2013
  text: 20131201
  day: 01
PublicationDecade 2010
PublicationTitle 2013 IEEE International Conference on Computer Vision
PublicationTitleAbbrev iccv
PublicationYear 2013
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0039286
ssj0001967680
Score 2.6273909
Snippet Optical flow computation is a key component in many computer vision systems designed for tasks such as action detection or activity recognition. However,...
SourceID unpaywall
proquest
ieee
SourceType Open Access Repository
Aggregation Database
Publisher
StartPage 1385
SubjectTerms Adaptive optics
Algorithms
Blends
Computer vision
Construction
deep convolutional networks
dense matching
Displacement
Energy management
Equations
Estimation
Integrated optics
large displacements
Matching
non-rigid matching
Nonlinear optics
Optical filters
optical flow
Optical imaging
Polymer blends
Sampling
SummonAdditionalLinks – databaseName: IEEE Electronic Library (IEL)
  dbid: RIE
  link: http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV3fT8IwEG6QF31CBSP-Sk1MfHGDsdGuxjeQoBH1QQxvy9pdI5EMIkyif729AkOND-5pS7qt63V3312_6xFyBl5dJCxmjscBnIAr5cSMc0cbiMSZlFLHlm1xz7r94HbQHBTIRZ4LAwCWfAYuntq1_GSsMgyV1Qy4NerUKNwNzsUiV2sdTxHMIOf6Sgsbs2-rPCICd9AHWu-vWbtptZ6R1OW7HtILbVWVHwBzM0sn8cc8Ho2-2ZpOifRWvVxQTF7dbCZd9flrA8f_fsY2qayz-uhjbq92SAHSXVJawlC6_MmnZXLVBph0RuP5Jb1DnjhtD6eWuoWPpQ8TG_2m2IBiFJdic9ozKh2DWRXS71w_tbrOssiC89IIvZnD0AnRCQ8DHWulE2A6UHXhJ6CDWCnRFFoyLn1hDHvoh5IrzUKBbE6VmAvh75FiOk5hn9BGIpvaa0itAnMIlDMIpkPjBUsJnqySMg5GNFnsoxEtx6FKTldDH5m5jQsWcQrjbBp5jInQqJzQr5LzXCb5_dZ_qYtoqNR7hPKMjDwP_n7JIdnCFgsiyhEpzt4yODZwYiZP7Dz6AqMjx8Y
  priority: 102
  providerName: IEEE
Title DeepFlow: Large Displacement Optical Flow with Deep Matching
URI https://ieeexplore.ieee.org/document/6751282
https://www.proquest.com/docview/1669883283
https://inria.hal.science/hal-00873592
UnpaywallVersion submittedVersion
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV3NS8MwFA-yHfSksonzY0QQPLVbv_IhXsQ5h-jcwcE8lSZN2HB0xbUO_evN67pNvWhPKbyQ8F54X_m9F4TOldPmMYmI5VClLJ9KaUWEUksbF4kSIYSOCrRFn_SG_v0oGJUA2aIWJjF8t8fG7ywtQMuMLWic5gXcqNoqgYukCqoO-4Prl03TzNZEyndAanm2A5jB4qmUH17jdp6k0ccimk6_GZDuLrpbLb3EjbzaeSZs-fmrK-Pfe9tD9U2RHh6szc8-2lJJDV11lEq709niEj8Axht3JvMCdgVJQPyUFplrDAQYMrAYyPGjUceQiKqjYff2-aZnlQ8kWGOXOZlFIIDQMWW-jrTUsSLal23uxUr7kZQ84FoQKjxujDLzmKBSE8YBiSlj88O9A1RJZok6RNiNRaAdV2jpm4-DjBQnmpkIVgjliAaqAUfDdNkDIzRxhrFsbgOdrTgcmnMJlw1Romb5PHQI4cyoC-Y10MWa9ev5RezR5iGILQSxhUZsR_-mPEY7MFziSk5QJXvL1anxDjLRLEr4muUB-QLwqr12
linkProvider Unpaywall
linkToHtml http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV1LT8MwDLYQHMaJx0CMZ5CQuNCxrl3SIG6DaYMNOAzErWpSRyCmbmIrCH49cfYCxIGeWilt0zi1PzufY4Aj9Csy5Qn3fIHohUJrL-FCeMZCJMGVUiZxbIsb3rwPrx5rjwtwMsuFQURHPsMynbq1_LSvcwqVnVpwa9WpVbhLNetViHG21jyiIrnFzpWpHraG39V5JAzukRc032HztFWvPxCtKyj7RDB0dVV-QMxCng2Sj_ek1_tmbRor0Jn2c0wyeSnnI1XWn7-2cPzvh6zCxjyvj93NLNYaLGC2DisTIMomv_mwCOcXiINGr_9-xtrEFGcXz0NH3qLHstuBi38zasAojsuoOetYpU7hrA24b1x2601vUmbBe6pG_sjj5IaYVEShSYw2KXIT6ooMUjRhorWsSaO4UIG0pj0KIiW04ZEkPqdO7YUMNmEx62e4Bayaqprxq8ro0B6SJI2Sm8j6wUqhr0pQpMGIB-OdNOLJOJTgcDr0sZ3dtGSRZNjPh7HPuYys0omCEhzPZDK733kwFRk_a_0WkzxjK8_tv19yAIVmt9OO262b6x1YptZjWsouLI5ec9yz4GKk9t2c-gJPTssX
linkToUnpaywall http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwjV1LSwMxEA7SHvSk0or1RQTBU7bdV7IRL2KtRbT2YKGeliSb0GLZLrZr0V9vZrutj4vuKQsTdpkJ881MvkwQOtNuiydUUOIyrUnAlCKCMkaMDZEYlVIaUbAterQ7CO6G4bAkyBZnYVKrd2dk484SAZp2TKBxmh9y62qrFDaSKqg66PWvnr-aZjbHSr0BU8t3XOAMFlel_IgaN_M0E-8LMZl8A5DONrpdfXrJG3lx8rl01Mevrox__9sOqn8d0sP9Nfzsog2d1tBlW-usM5kuLvA9cLxxezwraFdQBMSPWVG5xiCAoQKLQRw_WHcMhag6GnRunq67pLwggYy8yJ0TCgmESVgUGGGUSTQ1gWpxP9EmEErxkBtJmfS5BeXIjyRThkYcmJgqsS_c30OVdJrqfYS9RIbG9aRRgX042EhzaiKbwUqpXdlANdBonC17YMQ2z7DI5jXQ6UrDsV2XsNkgUj3NZ7FLKY-su4j8Bjpfq349v8g9WjwGs8Vgttia7eDfkodoC4ZLXskRqsxfc31so4O5PCmXxieFO7x1
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%3Abook&rft.genre=proceeding&rft.title=2013+IEEE+International+Conference+on+Computer+Vision&rft.atitle=DeepFlow%3A+Large+Displacement+Optical+Flow+with+Deep+Matching&rft.au=Weinzaepfel%2C+Philippe&rft.au=Revaud%2C+Jerome&rft.au=Harchaoui%2C+Zaid&rft.au=Schmid%2C+Cordelia&rft.date=2013-12-01&rft.pub=IEEE&rft.issn=1550-5499&rft.spage=1385&rft.epage=1392&rft_id=info:doi/10.1109%2FICCV.2013.175&rft.externalDocID=6751282
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1550-5499&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1550-5499&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1550-5499&client=summon