Matches in SemOpenAlex for { <https://semopenalex.org/work/W4214614183> ?p ?o ?g. }
- W4214614183 abstract "We present Multiscale Vision Transformers (MViT) for video and image recognition, by connecting the seminal idea of multiscale feature hierarchies with transformer models. Multiscale Transformers have several channel-resolution scale stages. Starting from the input resolution and a small channel dimension, the stages hierarchically expand the channel capacity while reducing the spatial resolution. This creates a multiscale pyramid of features with early layers operating at high spatial resolution to model simple low-level visual information, and deeper layers at spatially coarse, but complex, high-dimensional features. We evaluate this fundamental architectural prior for modeling the dense nature of visual signals for a variety of video recognition tasks where it outperforms concurrent vision transformers that rely on large scale external pre-training and are 5-10× more costly in computation and parameters. We further remove the temporal dimension and apply our model for image classification where it outperforms prior work on vision transformers. Code is available at: https://github.com/facebookresearch/SlowFast." @default.
- W4214614183 created "2022-03-02" @default.
- W4214614183 creator A5001594573 @default.
- W4214614183 creator A5002869474 @default.
- W4214614183 creator A5019393431 @default.
- W4214614183 creator A5022792966 @default.
- W4214614183 creator A5029760000 @default.
- W4214614183 creator A5036069974 @default.
- W4214614183 creator A5038469427 @default.
- W4214614183 date "2021-10-01" @default.
- W4214614183 modified "2023-10-16" @default.
- W4214614183 title "Multiscale Vision Transformers" @default.
- W4214614183 cites W1536680647 @default.
- W4214614183 cites W1677182931 @default.
- W4214614183 cites W1968245656 @default.
- W4214614183 cites W197865394 @default.
- W4214614183 cites W2022735534 @default.
- W4214614183 cites W2097117768 @default.
- W4214614183 cites W2108598243 @default.
- W4214614183 cites W2111624873 @default.
- W4214614183 cites W2147800946 @default.
- W4214614183 cites W2194775991 @default.
- W4214614183 cites W2342662179 @default.
- W4214614183 cites W2618799552 @default.
- W4214614183 cites W2625366777 @default.
- W4214614183 cites W2889326796 @default.
- W4214614183 cites W2955874753 @default.
- W4214614183 cites W2963091558 @default.
- W4214614183 cites W2963093735 @default.
- W4214614183 cites W2963150697 @default.
- W4214614183 cites W2963246338 @default.
- W4214614183 cites W2963524571 @default.
- W4214614183 cites W2963563276 @default.
- W4214614183 cites W2963722382 @default.
- W4214614183 cites W2964080601 @default.
- W4214614183 cites W2981165461 @default.
- W4214614183 cites W2981385151 @default.
- W4214614183 cites W2981413347 @default.
- W4214614183 cites W2983446232 @default.
- W4214614183 cites W2984287396 @default.
- W4214614183 cites W2988396473 @default.
- W4214614183 cites W2990152177 @default.
- W4214614183 cites W2990503944 @default.
- W4214614183 cites W2992308087 @default.
- W4214614183 cites W2998508940 @default.
- W4214614183 cites W3034429256 @default.
- W4214614183 cites W3034572008 @default.
- W4214614183 cites W3034885317 @default.
- W4214614183 cites W3035303837 @default.
- W4214614183 cites W3035452548 @default.
- W4214614183 cites W3035682985 @default.
- W4214614183 cites W3138516171 @default.
- W4214614183 cites W3174068320 @default.
- W4214614183 cites W3206471682 @default.
- W4214614183 cites W3210279979 @default.
- W4214614183 cites W4214612132 @default.
- W4214614183 cites W4231697575 @default.
- W4214614183 doi "https://doi.org/10.1109/iccv48922.2021.00675" @default.
- W4214614183 hasPublicationYear "2021" @default.
- W4214614183 type Work @default.
- W4214614183 citedByCount "309" @default.
- W4214614183 countsByYear W42146141832021 @default.
- W4214614183 countsByYear W42146141832022 @default.
- W4214614183 countsByYear W42146141832023 @default.
- W4214614183 crossrefType "proceedings-article" @default.
- W4214614183 hasAuthorship W4214614183A5001594573 @default.
- W4214614183 hasAuthorship W4214614183A5002869474 @default.
- W4214614183 hasAuthorship W4214614183A5019393431 @default.
- W4214614183 hasAuthorship W4214614183A5022792966 @default.
- W4214614183 hasAuthorship W4214614183A5029760000 @default.
- W4214614183 hasAuthorship W4214614183A5036069974 @default.
- W4214614183 hasAuthorship W4214614183A5038469427 @default.
- W4214614183 hasBestOaLocation W42146141832 @default.
- W4214614183 hasConcept C11413529 @default.
- W4214614183 hasConcept C119599485 @default.
- W4214614183 hasConcept C127313418 @default.
- W4214614183 hasConcept C127413603 @default.
- W4214614183 hasConcept C153180895 @default.
- W4214614183 hasConcept C154945302 @default.
- W4214614183 hasConcept C165801399 @default.
- W4214614183 hasConcept C205372480 @default.
- W4214614183 hasConcept C3020199158 @default.
- W4214614183 hasConcept C31972630 @default.
- W4214614183 hasConcept C41008148 @default.
- W4214614183 hasConcept C45374587 @default.
- W4214614183 hasConcept C62649853 @default.
- W4214614183 hasConcept C66322947 @default.
- W4214614183 hasConceptScore W4214614183C11413529 @default.
- W4214614183 hasConceptScore W4214614183C119599485 @default.
- W4214614183 hasConceptScore W4214614183C127313418 @default.
- W4214614183 hasConceptScore W4214614183C127413603 @default.
- W4214614183 hasConceptScore W4214614183C153180895 @default.
- W4214614183 hasConceptScore W4214614183C154945302 @default.
- W4214614183 hasConceptScore W4214614183C165801399 @default.
- W4214614183 hasConceptScore W4214614183C205372480 @default.
- W4214614183 hasConceptScore W4214614183C3020199158 @default.
- W4214614183 hasConceptScore W4214614183C31972630 @default.
- W4214614183 hasConceptScore W4214614183C41008148 @default.
- W4214614183 hasConceptScore W4214614183C45374587 @default.
- W4214614183 hasConceptScore W4214614183C62649853 @default.