Learning to Forecast and Refine Residual Motion for Image-to-Video Generation.
Long ZhaoXi PengYu TianMubbasir KapadiaDimitris N. MetaxasPublished in: ECCV (15) (2018)
Keyphrases
- static images
- input image
- learning algorithm
- object motion
- image data
- visual data
- key frames
- feature points
- successive frames
- video sequences
- image frames
- visual cues
- image features
- space time
- image segmentation
- image classification
- multiscale
- single image
- image representation
- camera movement
- motion estimation
- temporal domain
- visual perception
- single frame
- video signals
- normalized correlation
- input video
- image motion
- image retrieval
- optical flow
- high resolution
- low level
- moving camera
- spatial and temporal
- image quality