1 paper · 1 filter
Rungui Zhou, Chuanzhi Zhou, Yuk-Kit Hou +1
Video models commonly use uniform grids even though visual complexity varies substantially across space and time. We introduce OctVideo, which approximates a video clip with an oct…