OmniMAE: Single Model Masked Pretraining on Images and Videos

Open in new window