Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?

We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances-an aspect often overlooked in the literature in favor of the meta-learning paradigm. We introduce a transductive inference for a given query image, leveraging the statistics of its unl...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) s. 13974 - 13983
Hlavní autoři: Boudiaf, Malik, Kervadec, Hoel, Masud, Ziko Imtiaz, Piantanida, Pablo, Ayed, Ismail Ben, Dolz, Jose
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.06.2021
Témata:
ISSN:1063-6919
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances-an aspect often overlooked in the literature in favor of the meta-learning paradigm. We introduce a transductive inference for a given query image, leveraging the statistics of its unlabeled pixels, by optimizing a new loss containing three complementary terms: i) the cross-entropy on the labeled support pixels; ii) the Shannon entropy of the posteriors on the unlabeled query-image pixels; and iii) a global KL-divergence regularizer based on the proportion of the predicted foreground. As our inference uses a simple linear classifier of the extracted features, its computational load is comparable to inductive inference and can be used on top of any base training. Foregoing episodic training and using only standard cross-entropy training on the base classes, our inference yields competitive performances on standard benchmarks in the 1-shot scenarios. As the number of available shots increases, the gap in performances widens: on PASCAL-5 i , our method brings about 5% and 6% improvements over the state-of-the-art, in the 5- and 10-shot scenarios, respectively. Furthermore, we introduce a new setting that includes domain shifts, where the base and novel classes are drawn from different datasets. Our method achieves the best performances in this more realistic setting. Our code is freely available online: https://github.com/mboudiaf/RePRI-for-Few-Shot-Segmentation.
AbstractList We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances-an aspect often overlooked in the literature in favor of the meta-learning paradigm. We introduce a transductive inference for a given query image, leveraging the statistics of its unlabeled pixels, by optimizing a new loss containing three complementary terms: i) the cross-entropy on the labeled support pixels; ii) the Shannon entropy of the posteriors on the unlabeled query-image pixels; and iii) a global KL-divergence regularizer based on the proportion of the predicted foreground. As our inference uses a simple linear classifier of the extracted features, its computational load is comparable to inductive inference and can be used on top of any base training. Foregoing episodic training and using only standard cross-entropy training on the base classes, our inference yields competitive performances on standard benchmarks in the 1-shot scenarios. As the number of available shots increases, the gap in performances widens: on PASCAL-5 i , our method brings about 5% and 6% improvements over the state-of-the-art, in the 5- and 10-shot scenarios, respectively. Furthermore, we introduce a new setting that includes domain shifts, where the base and novel classes are drawn from different datasets. Our method achieves the best performances in this more realistic setting. Our code is freely available online: https://github.com/mboudiaf/RePRI-for-Few-Shot-Segmentation.
Author Ayed, Ismail Ben
Kervadec, Hoel
Dolz, Jose
Boudiaf, Malik
Masud, Ziko Imtiaz
Piantanida, Pablo
Author_xml – sequence: 1
  givenname: Malik
  surname: Boudiaf
  fullname: Boudiaf, Malik
  email: malik.boudiaf.1@etsmtl.net
  organization: ÉTS Montreal
– sequence: 2
  givenname: Hoel
  surname: Kervadec
  fullname: Kervadec, Hoel
  organization: ÉTS Montreal
– sequence: 3
  givenname: Ziko Imtiaz
  surname: Masud
  fullname: Masud, Ziko Imtiaz
  organization: ÉTS Montreal
– sequence: 4
  givenname: Pablo
  surname: Piantanida
  fullname: Piantanida, Pablo
  organization: Université Paris-Saclay,CentraleSupélec-CNRS
– sequence: 5
  givenname: Ismail Ben
  surname: Ayed
  fullname: Ayed, Ismail Ben
  organization: ÉTS Montreal
– sequence: 6
  givenname: Jose
  surname: Dolz
  fullname: Dolz, Jose
  organization: ÉTS Montreal
BookMark eNotzMtOAjEUgOFqNBGQJ9BFX2Cwp1fqxhAiSIKXCGqMC1I6pzAGWjJTNL69JLr6_9XXJicxRSTkElgPgNmr4evTs9RSmB5nHHoMhNFHpA1aKykVs_yYtIBpUWgL9ox0m-aTMSY4gLb9FvkY4XcxW6dMZ7jaYswuVynStyqv0z7Te8yumKKrYxVX13RAxymVdF672JR7n6svpJMYsMboD9fQwWZD39OePiCWN-fkNLhNg93_dsjL6HY-vCumj-PJcDAt1rwvciEdl9IHCEFzKP3S2aCtFShRGM-VCWZpgjMBwEOpZMkdD8oqwNDXzIul6JCLP7dCxMWurrau_llYZQw_OL-VvVUG
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IH
CBEJK
RIE
RIO
DOI 10.1109/CVPR46437.2021.01376
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
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
EISBN 1665445092
9781665445092
EISSN 1063-6919
EndPage 13983
ExternalDocumentID 9577299
Genre orig-research
GroupedDBID 6IE
6IH
6IL
6IN
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
IJVOP
OCL
RIE
RIL
RIO
ID FETCH-LOGICAL-h283t-4a244cf1ff621dcba9f6993e4e37c257f7b7fa7f11c1d54d2a2f5951ef860c3b3
IEDL.DBID RIE
ISICitedReferencesCount 148
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000742075004019&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:28:30 EDT 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-h283t-4a244cf1ff621dcba9f6993e4e37c257f7b7fa7f11c1d54d2a2f5951ef860c3b3
OpenAccessLink https://hal.science/hal-03979797
PageCount 10
ParticipantIDs ieee_primary_9577299
PublicationCentury 2000
PublicationDate 2021-June
PublicationDateYYYYMMDD 2021-06-01
PublicationDate_xml – month: 06
  year: 2021
  text: 2021-June
PublicationDecade 2020
PublicationTitle Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online)
PublicationTitleAbbrev CVPR
PublicationYear 2021
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0003211698
Score 2.6002963
Snippet We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances-an aspect often overlooked in the...
SourceID ieee
SourceType Publisher
StartPage 13974
SubjectTerms Benchmark testing
Codes
Computer vision
Entropy
Feature extraction
Image segmentation
Training
Title Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?
URI https://ieeexplore.ieee.org/document/9577299
WOSCitedRecordID wos000742075004019&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/eLvHCXMwlV3PT8IwFG6AePCECkbxR3rw6ICu3bp6MYSIkigh4g8SD2RrX4EENwND_33bbZkx8eKlaXp5SV_6vu-1_d5D6MKjwncDqRwvMgPTrhUrS-0ooErzIAIv9_Q9H42C6VSMK-iy1MIAQPb5DNp2mr3lq0Ru7VVZR3iWC4oqqnLu51qt8j6FmkzGF0GhjiNd0em_jB-ZfZcyWaBL2ra2nv-rh0oGIYP6_4zvoeaPFg-PS5TZRxWID1C9II-4OJqbBnobwJczWSQpnsD8vVAUxfh1mS6SbYofIA2dopjq_Ar38G2SKJwhla34amIeHpb2hhvcW62wiQR4ZOxeN9Hz4Oapf-cUnROchaELqcNCg9pSE619lygZhUL7hogAA8qlOaSaR1yHXBMiifKYckNXe4ZrgQ78rqQRPUS1OInhCGFFCUShR5U0iaDWIAz_YowLUCGThk8co4bdq9lHXhxjVmxT6-_lE7RrnZH_tTpFtXS9hTO0Iz_T5WZ9nnn0G39WosU
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PT8IwFG4QTfSECsbf9uDRAVu7H_ViCBEhwkIElcQD2dpXIMHNwNB_33ZbZky8eGmaXl7Sl77ve22_9xC6tglzLI8Lww7VQKWlxcpcGgKIkK4Xgp15uu_6vjeZsGEJ3RRaGABIP59BXU_Tt3wR842-KmswW3NBtoW2bUqtZqbWKm5UiMplHObl-jizyRrtl-ET1S9TKg-0zLquruf86qKSgkin8j_z-6j2o8bDwwJnDlAJokNUyekjzg_nuoreOvBljOZxgkcwe881RRF-XSTzeJPgASSBkZdTnd3iFn6IY4FTrNI1X1XUw73CXm-NW8slVrEA-8ruXQ09d-7H7a6R904w5oowJAYNFG5zaUrpWKbgYcCko6gIUCAuV8dUuqErA1eaJjeFTYUVWNJWbAuk5zQ5CckRKkdxBMcIC2JCGNhEcJUKSglMMTBKXQYioFwxihNU1Xs1_cjKY0zzbTr9e_kK7XbHg_603_Mfz9Cedkz28-oclZPVBi7QDv9MFuvVZerdbyWlpgw
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=proceeding&rft.title=Proceedings+%28IEEE+Computer+Society+Conference+on+Computer+Vision+and+Pattern+Recognition.+Online%29&rft.atitle=Few-Shot+Segmentation+Without+Meta-Learning%3A+A+Good+Transductive+Inference+Is+All+You+Need%3F&rft.au=Boudiaf%2C+Malik&rft.au=Kervadec%2C+Hoel&rft.au=Masud%2C+Ziko+Imtiaz&rft.au=Piantanida%2C+Pablo&rft.date=2021-06-01&rft.pub=IEEE&rft.eissn=1063-6919&rft.spage=13974&rft.epage=13983&rft_id=info:doi/10.1109%2FCVPR46437.2021.01376&rft.externalDocID=9577299