AI Paper Atlas

Research map / Video / Action recognition

LARY: A Latent Action Representation Yielding Benchmark

Dujun Nie, Fengjiao Chen, Jun Kuang, Qi Lv, Xiaoyu Li, Xuezhi Cao

ECCV 2026 · Video › Action recognition · approach: Action anticipation and latent actions · 17 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 general latent-feature LAM (LAPA-DINOv2) instead of embodied UniVLALAPA framework · LARYBench classificationAvg accuracy: 18.82 -> 57.41Improved
use latent-based V-JEPA 2.1 instead of pixel-based Wan2.2encoder · LARYBench classificationAvg accuracy: 66.58 -> 76.29Improved
replace SigLIP2 or MAGVIT2 encoder with DINOv3 in LAPALAPA General LAM · LARYBench classificationAvg accuracy: 42.09 -> 62.73Improved
use DINOv3 encoder directly instead of LAPA-DINOv3General LAM · LARYBench absolute regression CALVINnormalized MSE: 0.50 -> 0.18Improved
use best General LAM (LAPA-DINOv3) in place of V-JEPA 2.1 for delta regressiondelta regression · LARYBench delta regressionAvg normalized MSE: 0.50 -> 0.57No clear effect
increase codebook size 8 to 64 (ablation)LAPA-DINOv3 · Composite Humanaccuracy: 71.31 -> 70.15No clear effect
increase codebook size 64 to 256 (ablation)LAPA-DINOv3 · Composite Humancodebook utilization: 100.0 -> 89.5Hurt
increase sequence length 16 to 49 (ablation)LAPA-DINOv3 · codebook utilizationutilization %: 1.6 -> 100.0Improved
increase sequence length 49 to 64 (ablation)LAPA-DINOv3 · AgiBotWorld-Beta regressionMSE: 0.83 -> 0.72Mixed
increase latent dimension 32 to 256 (ablation)LAPA-DINOv3 · Composite Humanaccuracy: 60.94 -> 70.15Improved
increase latent dimension 256 to 1024 (ablation)LAPA-DINOv3 · Composite Humanaccuracy: 70.15 -> 72.55Mixed
increase sampling stride 5 to 30FLUX.2-dev · VLABench regressionMSE: 0.04 -> 0.62Hurt

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