A 12.19M-parameter adapter on a frozen video/tracking model forecasts future 3D object-point tracks from 7 observed frames, trained on 40k human videos without language or action labels.
hub
Rational imitation in preverbal infants
1 Pith paper cite this work, alongside 1,401 external citations. Polarity classification is still indexing.
1
Pith paper citing it
1,401
external citations · OpenAlex
hub tools
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MotionForesight: Re-purposing Video Models for Future 3D Scene-Flow Prediction
A 12.19M-parameter adapter on a frozen video/tracking model forecasts future 3D object-point tracks from 7 observed frames, trained on 40k human videos without language or action labels.