Reconstructing Spatial Aspects of Motion by Image-to-Path Deep Neural Networks.
Rok PahicAndrej GamsAles UdePublished in: IEEE Robotics Autom. Lett. (2021)
Keyphrases
- neural network
- spatial information
- static images
- spatial and temporal
- spatial distribution
- single image
- image features
- image data
- temporal continuity
- input image
- visual perception
- image content
- image retrieval
- image classification
- feature points
- template matching
- image segmentation
- motion analysis
- spatial relationships
- space time
- segmentation method
- edge detection
- spatial relations
- multiscale
- spatial correlation
- spatial frequency
- image representation
- motion estimation
- high resolution
- image frames
- spatio temporal
- image sequences
- vector field
- color distribution
- image motion
- temporal domain
- spatial arrangement
- motion model
- image regions
- optical flow
- ego motion
- ground plane
- object motion
- reference frame
- key frames
- action recognition
- camera movement
- displacement estimation