Transferring Physical Motion Between Domains for Neural Inertial Tracking.
Changhao ChenYishu MiaoChris Xiaoxuan LuPhil BlunsomAndrew MarkhamNiki TrigoniPublished in: CoRR (2018)
Keyphrases
- motion model
- physically plausible
- motion analysis
- motion tracking
- robust tracking
- motion segmentation
- motion detection
- kalman filter
- articulated body
- deformable objects
- tracking multiple
- real world
- image sequences
- transfer learning
- tracking framework
- real time tracking
- abrupt motion
- motion estimation
- moving target
- neural network
- particle filter
- motion parameters
- rigid objects
- upper body
- extended kalman filters
- moving camera
- articulated motion
- multiple moving objects
- consecutive frames
- object motion
- nonrigid motion
- optical flow
- visual tracking
- space time
- motion vector fields
- network architecture
- moving objects
- motion field
- temporal continuity
- head motion
- real sequences
- color based tracking
- hand motion
- object tracking
- moving platform
- fixed camera
- inertial sensors
- markerless
- particle filtering
- camera motion
- human walking
- wide area motion imagery
- articulated objects
- dynamic model
- angular velocity
- computer vision
- point correspondences
- physical constraints
- motion sequences
- beating heart
- motion capture
- reference frame
- mobile robot
- human gait
- human body
- appearance model