MaskUKF: An Instance Segmentation Aided Unscented Kalman Filter for 6D Object Pose and Velocity Tracking.
Nicola A. PigaFabrizio BottarelClaudio FantacciGiulia VezzaniUgo PattaciniLorenzo NatalePublished in: Frontiers Robotics AI (2021)
Keyphrases
- d objects
- unscented kalman filter
- position and orientation
- pose estimation
- constant velocity
- kalman filter
- physically plausible
- state estimation
- multi view
- viewpoint
- dynamic model
- motion model
- visual tracking
- object recognition
- extended kalman filter
- inertial sensors
- motion segmentation
- joint angles
- three dimensional
- image segmentation
- particle filter
- particle filtering
- range data
- multiple views
- markerless
- multiple objects
- kalman filtering
- simultaneous localization and mapping
- inverse kinematics
- stereo camera
- human pose
- appearance model
- object tracking
- motion tracking
- stereo pair
- optical flow