Learning by Aligning Videos in Time (CVPR 2021)

A Self-Supervised Approach for Training Viewpoint-, Actor-, and Scene-Invariant Video Representations.

DOWNLOAD THE OPEN ACCESS VERSION FROM THE COMPUTER VISION FOUNDATION (CVF)
Get the code of official implementation of our CVPR 2021 paper

DENSE PER-FRAME LABELS FOR 2123 VIDEOS ANNOTATED BY US.