AI Paper Atlas

Research map / Video / Video-language understanding

Principles of Visual Tokens for Efficient Video Understanding

Xinyue Hao, Gen Li, Shreyank N Gowda, Robert B. Fisher, Jonathan Huang, Anurag Arnab, Laura Sevilla-Lara

ICCV 2025 · Video › Video-language understanding · approach: Long video understanding · 5 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
compute oracle gradient with the true label instead of the predicted label (ablation)token oracle, 50% tokens · SS-V2Top-1 accuracy: 70.00 -> 78.52Improved
add adaptive budget (LITE++) to LITE90 (ablation)VideoMAE-LITE · SS-V2GFLOPs / Top-1: 159x2x3 -> 105x2x3 GFLOPs; 70.54 -> 69.94Mixed
add adaptive budget (LITE++) to LITE70 (ablation)VideoMAE-LITE · SS-V2GFLOPs / Top-1: 118x2x3 -> 84x2x3 GFLOPs; 70.44 -> 69.86Mixed
add adaptive budget (LITE++) to LITE30 (ablation)VideoMAE-LITE · SS-V2GFLOPs / Top-1: 46x2x3 -> 40x2x3 GFLOPs; 68.33 -> 67.74Mixed
replace VideoMAE backbone with weaker TimeSformer for LITE50LITE · SS-V2Top-1 accuracy: 56.23 -> 48.28 (vs full TimeSformer-B)Hurt
select tokens with LITE90 instead of using all tokensVideoMAE · SS-V2Top-1 accuracy: 70.50 -> 70.54No clear effect
select tokens with LITE50 instead of using all tokensVideoMAE · Kinetics-400Top-1 accuracy: 81.25 -> 80.36Hurt
use LITE instead of ToMe at similar GFLOPsVideoMAE · SS-V2Top-1 accuracy: 69.63 -> 69.91Improved
use LITE instead of ToMe at similar GFLOPsVideoMAE · Kinetics-400Top-1 accuracy: 79.95 -> 80.36Improved
transfer selector trained on K400 to SS-V2 (LITE90)VideoMAE-LITE · SS-V2Top-1 accuracy: 70.50 -> 70.28No clear effect
transfer selector trained on K400 to AVA detection (LITE70)VideoMAE-LITE · AVAmAP: 31.50 -> 30.83No clear effect
use a more complex selector (transformer blocks, more layers, global information, improved oracle) (ablation)LITE selector · SS-V2Top-1 accuracyNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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