3D Pose from Motion for Cross-View Action Recognition via Non-linear Circulant Temporal Encoding.
Ankur GuptaJulieta MartinezJames J. LittleRobert J. WoodhamPublished in: CVPR (2014)
Keyphrases
- space time
- spatial and temporal
- cross view action recognition
- temporal consistency
- motion analysis
- motion estimation
- image sequences
- motion model
- spatio temporal
- position and orientation
- camera motion
- least squares
- motion patterns
- human motion
- multi view
- optical flow
- pose estimation
- view invariant
- motion capture
- feature space
- temporal continuity
- moving objects
- d objects
- dynamic textures
- motion trajectories
- fourier transform
- feature points
- humanoid robot
- dynamic scenes
- articulated objects
- three dimensional