Learn the Force We Can: Multi-Object Video Generation from Pixel-Level Interactions.
Aram DavtyanPaolo FavaroPublished in: CoRR (2023)
Keyphrases
- pixel level
- multi object
- multiple objects
- superpixels
- video frames
- foreground objects
- video sequences
- motion trajectories
- video data
- image fusion
- human perception
- multimedia
- motion analysis
- three dimensional
- face recognition
- high quality
- multiscale
- dynamic scenes
- image classification
- medical images
- higher order
- motion estimation