VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking

Open in new window