Skip to content
arXiv cs.CV · Papers

Fre-Res: Frequency-Residual Video Token Compression for Efficient Video MLLMs

arXiv:2605.16366v2 Announce Type: replace Abstract: Video MLLMs face a persistent tension between spatial fidelity and temporal coverage: preserving fine-grained visual details requires many spatial tokens, while capturing short-lived events requires dense temporal sampling. We propose textbf{Fre-Res}, a budget-adapti