MoVA introduces modular asymmetric dual projections to handle temporal misalignment and semantic asymmetry in long video-text alignment.
arXiv preprint arXiv:2202.04828 (2022)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MoVA: Learning Asymmetric Dual Projections for Modular Long Video-Text Alignment
MoVA introduces modular asymmetric dual projections to handle temporal misalignment and semantic asymmetry in long video-text alignment.