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

Zhan Tong Yibing Song Jue Wang Limin Wang

Published in: CoRR (2022)

Keyphrases

synthetic data
data sets
raw data
data analysis
data sources
training samples
data collection
image data
high quality
training data
database
real time
data processing
training examples
visual data
multimedia data
sensor data
denoising
small number
video sequences
neural network