Where2Act: From Pixels to Actions for Articulated 3D Objects.
Kaichun MoLeonidas J. GuibasMustafa MukadamAbhinav GuptaShubham TulsianiPublished in: ICCV (2021)
Keyphrases
- d objects
- object recognition
- viewpoint
- pose estimation
- multi view
- three dimensional
- human body
- range data
- shape descriptors
- line drawings
- cad models
- invariant recognition
- d mesh
- body parts
- curved surfaces
- polyhedral objects
- object classes
- texture mapping
- pose normalization
- machine learning
- human actions
- object model
- input image
- action sequences
- topological information
- feature vectors