The Role of ImageNet Classes in Fréchet Inception Distance
Fréchet Inception Distance (FID) is the primary metric for ranking models in data-driven generative modeling. While remarkably successful, the metric is known to sometimes disagree with human judgement. We investigate a root cause of these discrepancies, and visualize what FID "looks at" i...
Saved in:
Published in | arXiv.org |
---|---|
Main Authors | , , , , |
Format | Paper |
Language | English |
Published |
Ithaca
Cornell University Library, arXiv.org
14.02.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | Fréchet Inception Distance (FID) is the primary metric for ranking models in data-driven generative modeling. While remarkably successful, the metric is known to sometimes disagree with human judgement. We investigate a root cause of these discrepancies, and visualize what FID "looks at" in generated images. We show that the feature space that FID is (typically) computed in is so close to the ImageNet classifications that aligning the histograms of Top-\(N\) classifications between sets of generated and real images can reduce FID substantially -- without actually improving the quality of results. Thus, we conclude that FID is prone to intentional or accidental distortions. As a practical example of an accidental distortion, we discuss a case where an ImageNet pre-trained FastGAN achieves a FID comparable to StyleGAN2, while being worse in terms of human evaluation. |
---|---|
AbstractList | Fréchet Inception Distance (FID) is the primary metric for ranking models in data-driven generative modeling. While remarkably successful, the metric is known to sometimes disagree with human judgement. We investigate a root cause of these discrepancies, and visualize what FID "looks at" in generated images. We show that the feature space that FID is (typically) computed in is so close to the ImageNet classifications that aligning the histograms of Top-\(N\) classifications between sets of generated and real images can reduce FID substantially -- without actually improving the quality of results. Thus, we conclude that FID is prone to intentional or accidental distortions. As a practical example of an accidental distortion, we discuss a case where an ImageNet pre-trained FastGAN achieves a FID comparable to StyleGAN2, while being worse in terms of human evaluation. |
Author | Karras, Tero Aila, Timo Kynkäänniemi, Tuomas Lehtinen, Jaakko Aittala, Miika |
Author_xml | – sequence: 1 givenname: Tuomas surname: Kynkäänniemi fullname: Kynkäänniemi, Tuomas – sequence: 2 givenname: Tero surname: Karras fullname: Karras, Tero – sequence: 3 givenname: Miika surname: Aittala fullname: Aittala, Miika – sequence: 4 givenname: Timo surname: Aila fullname: Aila, Timo – sequence: 5 givenname: Jaakko surname: Lehtinen fullname: Lehtinen, Jaakko |
BookMark | eNrjYmDJy89LZWLgNDI2NtS1MDEy4mDgLS7OMjAwMDIzNzI1NeZksA7JSFUIys9JVchPU_DMTUxP9UstUXDOSSwuTi1WyMxTcCs6vDI5AyjmmZecWlCSmZ-n4JJZXJII5PEwsKYl5hSn8kJpbgZlN9cQZw_dgqL8wtLU4pL4rPzSojygVLyRmbGFham5kbGhMXGqAAlDOKs |
ContentType | Paper |
Copyright | 2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License. |
Copyright_xml | – notice: 2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License. |
DBID | 8FE 8FG ABJCF ABUWG AFKRA AZQEC BENPR BGLVJ CCPQU DWQXO HCIFZ L6V M7S PIMPY PQEST PQQKQ PQUKI PRINS PTHSS |
DatabaseName | ProQuest SciTech Collection ProQuest Technology Collection Materials Science & Engineering Collection ProQuest Central (Alumni) ProQuest Central ProQuest Central Essentials ProQuest Central Technology Collection ProQuest One Community College ProQuest Central SciTech Premium Collection ProQuest Engineering Collection Engineering Database Publicly Available Content Database ProQuest One Academic Eastern Edition (DO NOT USE) ProQuest One Academic ProQuest One Academic UKI Edition ProQuest Central China Engineering Collection |
DatabaseTitle | Publicly Available Content Database Engineering Database Technology Collection ProQuest Central Essentials ProQuest One Academic Eastern Edition ProQuest Central (Alumni Edition) SciTech Premium Collection ProQuest One Community College ProQuest Technology Collection ProQuest SciTech Collection ProQuest Central China ProQuest Central ProQuest Engineering Collection ProQuest One Academic UKI Edition ProQuest Central Korea Materials Science & Engineering Collection ProQuest One Academic Engineering Collection |
DatabaseTitleList | Publicly Available Content Database |
Database_xml | – sequence: 1 dbid: 8FG name: ProQuest Technology Collection url: https://search.proquest.com/technologycollection1 sourceTypes: Aggregation Database |
DeliveryMethod | fulltext_linktorsrc |
Discipline | Physics |
EISSN | 2331-8422 |
Genre | Working Paper/Pre-Print |
GroupedDBID | 8FE 8FG ABJCF ABUWG AFKRA ALMA_UNASSIGNED_HOLDINGS AZQEC BENPR BGLVJ CCPQU DWQXO FRJ HCIFZ L6V M7S M~E PIMPY PQEST PQQKQ PQUKI PRINS PTHSS |
ID | FETCH-proquest_journals_26388572313 |
IEDL.DBID | BENPR |
IngestDate | Thu Oct 10 19:36:25 EDT 2024 |
IsOpenAccess | true |
IsPeerReviewed | false |
IsScholarly | false |
Language | English |
LinkModel | DirectLink |
MergedId | FETCHMERGED-proquest_journals_26388572313 |
OpenAccessLink | https://www.proquest.com/docview/2638857231?pq-origsite=%requestingapplication% |
PQID | 2638857231 |
PQPubID | 2050157 |
ParticipantIDs | proquest_journals_2638857231 |
PublicationCentury | 2000 |
PublicationDate | 20230214 |
PublicationDateYYYYMMDD | 2023-02-14 |
PublicationDate_xml | – month: 02 year: 2023 text: 20230214 day: 14 |
PublicationDecade | 2020 |
PublicationPlace | Ithaca |
PublicationPlace_xml | – name: Ithaca |
PublicationTitle | arXiv.org |
PublicationYear | 2023 |
Publisher | Cornell University Library, arXiv.org |
Publisher_xml | – name: Cornell University Library, arXiv.org |
SSID | ssj0002672553 |
Score | 3.4569743 |
SecondaryResourceType | preprint |
Snippet | Fréchet Inception Distance (FID) is the primary metric for ranking models in data-driven generative modeling. While remarkably successful, the metric is known... |
SourceID | proquest |
SourceType | Aggregation Database |
SubjectTerms | Affine transformations Histograms Seat belts |
Title | The Role of ImageNet Classes in Fréchet Inception Distance |
URI | https://www.proquest.com/docview/2638857231 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwY2BQMTE1MEk1NUrVNTJLS9Q1MU-z1E1KSrLQTTFOSUo2TAK2gRNBHUVfPzOPUBOvCNMI6IBbMXRZJaxMBBfUKfnJoDFyfSNgQrEwNQc2R-wLCnVBt0aBZlehV2gwM7AaGZqApmlZnVz9AoLgoyxGZubANrMxRkELrj3cBBlYAxILUouEGJhS84QZ2MGLLpOLRRisgXGkEJSfk6qQn6bgmQvM2X6pJQrgaypTixUy8xTcig6vBAZqiYJnHnT1iYILqL0H5IkyKLu5hjh76MJsjIemiuJ4hB-MxRhYgN37VAkGBSNLgxSDNFND82SjVJOUpJQkYL8uLSUNdJKapWFiqpkkgww-k6TwS0szcIEuSAetMzY0kWFgKSkqTZUFVqMlSXIMzBZu7nLQEAPyfOtcAaU0fVM |
link.rule.ids | 786,790,12792,21416,33406,33777,43633,43838 |
linkProvider | ProQuest |
linkToHtml | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV1LS8QwEB50F9GbT3ysGtBrsE3TFx48qLXV3SKywt5K00xgQdu1rT_K3-EfMyldPQh7DIGEJJOZbyYf-QAuuWtxdBlS5qmccl-FVAgRUOlIUdhCY-DcJIqT1Itf-ePMnfUFt6anVS59YueoZVWYGvkV04YSuL6GIzeLD2pUo8zrai-hsQ5D7niOsfMgevitsTDP14jZ-edmu9gRbcPwOV9gvQNrWO7CRke5LJo9uNYnRF6qNySVIsm7vtcptqQTqcSGzEsS1d9fektbkpQ994TcGbSnW_twEd1Pb2O6nDHrbaLJ_lbgHMBAJ_d4CISFlrSUa_sFQy6FFDqrU1KZf9RCO0fvCEarRjpe3X0Om_F0Ms7GSfp0AltGKt0wjm0-gkFbf-KpDqitOOt27QehGHzD |
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=The+Role+of+ImageNet+Classes+in+Fr%C3%A9chet+Inception+Distance&rft.jtitle=arXiv.org&rft.au=Kynk%C3%A4%C3%A4nniemi%2C+Tuomas&rft.au=Karras%2C+Tero&rft.au=Aittala%2C+Miika&rft.au=Aila%2C+Timo&rft.date=2023-02-14&rft.pub=Cornell+University+Library%2C+arXiv.org&rft.eissn=2331-8422 |