vid-TLDR: Training Free Token merging for Light-Weight Video Transformer.

Joonmyung Choi, Sanghyeok Lee, Jaewon Chu, Minhyuk Choi, Hyunwoo J. Kim

Browse the full CVPR paper archive.