Is ImageNet worth 1 video? Learning strong image encoders from 1 long unlabelled video.
Shashanka VenkataramananMamshad Nayeem RizveJoão CarreiraYuki M. AsanoYannis AvrithisPublished in: CoRR (2023)
Keyphrases
- video sequences
- video images
- weakly labeled
- image frames
- image collections
- learning algorithm
- image features
- input image
- visual data
- video streams
- learning process
- video data
- image classification
- video frames
- segmentation method
- single image
- low level
- multiscale
- multimedia
- object motion
- image content
- static images
- temporal continuity
- video clips
- video surveillance
- video content
- image representation
- image data
- real time video
- key frames
- video analysis
- semi supervised learning
- visual concepts
- image quality
- compressed video
- active learning
- image retrieval
- image segmentation
- video files