Where2Act: From Pixels to Actions for Articulated 3D Objects.
Kaichun MoLeonidas J. GuibasMustafa MukadamAbhinav GuptaShubham TulsianiPublished in: CoRR (2021)
Keyphrases
- d objects
- viewpoint
- object recognition
- multi view
- pose estimation
- range data
- three dimensional
- shape descriptors
- cad models
- input image
- object features
- image contours
- d mesh
- object classes
- invariant recognition
- human body
- line drawings
- curved surfaces
- topological information
- object model
- geometric models
- real objects
- texture mapping
- human actions
- object representation
- motion capture
- computer generated images