AI Paper Atlas

Research map / Video / Action recognition

MultiModal Action Conditioned Video Simulation

Yichen Li, Antonio Torralba

ICCV 2025 · Video › Action recognition · approach: Action anticipation and latent actions · 1 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 multisensory actions instead of UniSim sentence textI2VGen simulator · ActionSenseFVD: 251.7 -> 203.5Improved
combine all four modalities instead of hand-pose only (ablation)multisensory simulator · ActionSenseLPIPS: 0.349 -> 0.276Improved
combine all modalities instead of muscle EMG only (ablation)multisensory simulator · ActionSenseFVD: 348.2 -> 203.5Improved
add text phrase to multisensory conditioningmultisensory simulator · ActionSenseFVD: 203.5 -> 200.4No clear effect
use our feature instead of ImageBindmultimodal feature · ActionSenseFVD: 315.6 -> 203.5Improved
use our feature instead of SignalAgnosticmultimodal feature · ActionSenseLPIPS: 0.361 -> 0.276Improved
use softmax fusion instead of max pooling (ablation)multimodal fusion · ActionSenseFVD: 284.8 -> 203.5Improved
use softmax fusion instead of concatenation (ablation)multimodal fusion · ActionSenseFVD: 279.9 -> 203.5Improved
add interaction feature y' regularization (ablation)multisensory simulator · ActionSenseFVD: 339.0 -> 203.5Improved
use relaxed hyperplane instead of hard projection (ablation)interaction feature · ActionSenseFVD: 265.5 -> 203.5Improved
train on all modalities, test with one modality removed (ablation)multisensory simulator · ActionSenseFVD: 203.5 -> 205.6No clear effect
add simulator loss Lsim to policy trainingdiffusion policy · goal-conditioned policypolicy MSE: 1.01 -> 0.90Improved

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