S.T.A.R.-Track: Latent Motion Models for End-to-End 3D Object Tracking With Adaptive Spatio-Temporal Appearance Representations.
Simon DollNiklas HanselmannLukas SchneiderRichard SchulzMarkus EnzweilerHendrik P. A. LenschPublished in: IEEE Robotics Autom. Lett. (2024)
Keyphrases
- end to end
- motion model
- d objects
- spatio temporal
- image sequences
- object representation
- particle filter
- motion segmentation
- motion patterns
- optical flow
- motion analysis
- scalable video
- appearance model
- motion estimation
- object recognition
- three dimensional
- pose estimation
- motion field
- viewpoint
- object tracking
- multi view
- tracking framework
- motion parameters
- shape descriptors
- range data
- object parts
- multiple views
- image motion
- moving objects
- visual tracking
- feature tracking
- kalman filter
- space time
- feature points
- dynamic scenes
- structure from motion
- computer vision