VideoMAE: Masked Autoencoders Are Data-Efficient Learners for Self-Supervised Video Pre-Training

Tong, Zhan, Song, Yibing, Wang, Jue · Advances in Neural Information Processing Systems 35 · 2022