ASCNet: Self-supervised Video Representation Learning with Appearance-Speed Consistency

Published in ICCV, 2020