RayTran: 3D pose estimation and shape reconstruction of multiple objects from videos with ray-traced transformers.
Michal J. TyszkiewiczKevis-Kokitsi ManinisStefan PopovVittorio FerrariPublished in: CoRR (2022)
Keyphrases
- multiple objects
- shape reconstruction
- photometric stereo
- shape from shading
- shape recognition
- multiple images
- real objects
- video sequences
- pose estimation
- particle filter
- reconstruction method
- complex scenes
- multiple cameras
- video data
- multi view
- cast shadows
- video surveillance
- light source
- single image
- high quality
- ground plane
- semi supervised
- pairwise