AI Paper Atlas

Research map / Video / Action recognition

How Can Objects Help Action Recognition?

Xingyi Zhou, Anurag Arnab, Chen Sun, Cordelia Schmid

CVPR 2023 · Video › Action recognition · approach: Disentangled action recognition · 19 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
Add object-aware attention while using all tokensVideoMAE-based ViViT · SomethingElse, SSv2, and Epic-KitchensTop-1 accuracy: +2.1, +1.3, and +0.6 percentage pointsImproved
Use object-guided token sampling instead of uniform random token samplingSpace-time ViViT · SomethingElse, SSv2, and Epic-KitchensTop-1 accuracyImproved
Drop background tokens guided by object detectionsSpace-time ViViT · SomethingElse, SSv2, and Epic-KitchensTop-1 accuracy: 0.2 pointsImproved
Use object-guided sampling with ground-truth detectionsSpace-time ViViT · SomethingElse and SSv2Top-1 accuracy: 2.1% gain on both Something datasetsImproved
Retain object-guided tokens and add two-view testing at 50% tokensObjectViViT · SomethingElse, SSv2, and Epic-KitchensTop-1 accuracyImproved
Retain 30%, 40%, and 60% of tokens, respectivelyObjectViViT · SomethingElse, SSv2, and Epic-KitchensTop-1 accuracy: 30%, 40%, and 60% of tokensNo clear effect
Retain only foreground/object tokens and remove all background tokens (ablation)Object-guided token sampling · SSv2Top-1 accuracyHurt
Apply token sampling before transformer layers rather than at layer 1 or layer 6 (ablation)Object-guided token sampling · SSv2Top-1 accuracy: 66.1 vs 66.0 and 65.9No clear effect
Use early token sampling rather than sampling at layer 1 or 6 (ablation)Object-guided token sampling · SSv2gFLOPs: 80 vs 88 and 130Improved
Replace RoIAlign or binary block-mask pooling with center-heatmap-weighted pooling (ablation)Object-aware attention · SSv2Top-1 accuracy: 67.2 -> 67.4Improved
Add tracking identity information rather than using no tracking (ablation)Object-aware attention · SSv2Top-1 accuracy: 67.0 -> 67.4Improved
Use identity attention rather than a shared identity embedding (ablation)Object-aware attention · SSv2Top-1 accuracy: 67.4 vs 67.4No clear effect

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