Hyper-VolTran: Fast and Generalizable One-Shot Image to 3D Object Structure via HyperNetworks.
Christian SimonSen HeJuan-Manuel Pérez-RúaMengmeng XuAmine BenhalloumTao XiangPublished in: CoRR (2023)
Keyphrases
- d objects
- physically plausible
- image contours
- input image
- image features
- object features
- image data
- multi view
- object recognition
- viewpoint
- invariant recognition
- pose estimation
- object matching
- three dimensional objects
- d mesh
- real world objects
- multiple views
- curved objects
- single image
- object representation
- multi views
- image content
- image retrieval
- object classes
- shape descriptors
- surface properties
- cad models
- image segmentation
- segmentation method
- recognition and pose estimation
- image synthesis
- computer vision algorithms
- surface patches
- test images
- image matching
- geometric invariants
- feature points