AI Paper Atlas

Research map / Video / Action recognition

ViViT: A Video Vision Transformer

Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lučić, Cordelia Schmid

ICCV 2021 · Video › Action recognition · approach: Contrastive video representation learning · 3,115 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use tubelet embedding with central frame init instead of filter inflation (ablation)ViViT-B Model 1 · Kinetics 400top-1 accuracy: 77.6 -> 79.2Improved
use tubelet embedding with central frame init instead of uniform frame sampling (ablation)ViViT-B Model 1 · Kinetics 400top-1 accuracy: 78.5 -> 79.2Improved
initialise tubelet embedding from pretrained image model rather than randomly (ablation)ViViT-B Model 1 · Kinetics 400top-1 accuracy: 73.2 -> 79.2Improved
add a temporal transformer (L_t=4) over frame representations instead of average pooling (ablation)ViViT-B Model 2 · Kinetics 400top-1 accuracy: 75.8 -> 78.8Improved
add a temporal transformer instead of average pooling (ablation)ViViT-B Model 2 · Epic Kitchensaction accuracy: 38.8 -> 43.7Improved
use unfactorised spatio-temporal attention instead of factorised encoder (ablation)ViViT-B · Kinetics 400top-1 accuracy: 78.8 -> 80.0Improved
use factorised encoder instead of unfactorised attention on a small dataset (ablation)ViViT-B · Epic Kitchensaction accuracy: 43.1 -> 43.7Improved
increase number of temporal transformer layers L_t from 1 to 12 (ablation)ViViT Model 2 · Kinetics 400top-1 accuracy: 78.6 -> 78.9No clear effect
add progressive regularisation (stochastic depth, random augment, label smoothing, mixup, K400 init) (ablation)ViViT Factorised encoder · Epic Kitchenstop-1 accuracy: 38.4 -> 43.7Improved
add the same regularisation set (ablation)ViViT Factorised encoder · Something-Something v2top-1 accuracy: +5%Improved
add label smoothing (ablation)ViViT Factorised encoder · Epic Kitchenstop-1 accuracy: 41.1 -> 43.1Improved
increase crop size from 224 to 320 (ablation)ViViT-L/16x2 Model 1 · Kinetics 400accuracy: 80.3 -> 81.0Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy