Video PreTraining (VPT): Learning to Act by Watching Unlabeled Online Videos.
Bowen BakerIlge AkkayaPeter ZhokovJoost HuizingaJie TangAdrien EcoffetBrandon HoughtonRaul SampedroJeff ClunePublished in: NeurIPS (2022)
Keyphrases
- online learning
- video data
- video sequences
- video content
- learning algorithm
- video frames
- online video
- active learning
- supervised learning
- real time
- unsupervised learning
- learning process
- multimedia
- video clips
- reinforcement learning
- video database
- high definition
- surveillance videos
- video event
- video analysis
- video surveillance
- key frames
- prior knowledge
- spatio temporal
- machine learning