#motion modeling

try —

2 papers match

cs.CV2026

The TIME Machine: On The Power of Motion for Efficient Perception

Mantas Skackauskas, Xinyue Hao, Laura Sevilla-Lara

The paper introduces TIME, a video representation learned from motion point-tracks using a masked autoencoder, enabling self‑supervised, language‑free training that requires far le…

#video representation#motion modeling#self-supervised learning#masked autoencoder

cs.CV2026

ScanFocus: A Coarse-to-Fine Framework for Spatio-Temporal Video Grounding

Kai Chen, Ming Dai, Wenxuan Cheng +1

The paper introduces ScanFocus, a coarse-to-fine framework for spatio-temporal video grounding that first scans videos globally to generate coarse object proposals and then refines…

#spatio-temporal video grounding#vision-language fusion#coarse-to-fine modeling#temporal localization