RandomRooms: Unsupervised Pre-training from Synthetic Shapes and Randomized Layouts for 3D Object Detection.
Yongming RaoBenlin LiuYi WeiJiwen LuCho-Jui HsiehJie ZhouPublished in: ICCV (2021)
Keyphrases
- d objects
- shape descriptors
- line drawings
- image contours
- object recognition
- viewpoint
- partial matching
- supervised learning
- multi view
- shape matching
- multiple views
- pose estimation
- three dimensional objects
- geometric models
- range data
- topological information
- three dimensional
- object contours
- d mesh
- object detection
- physically plausible
- semi supervised
- training set
- surface patches
- object classes
- shape analysis
- planar objects
- object detectors
- minimum bounding
- object representation
- cad models
- shape representation
- detection method
- shape prior
- surface features
- real world objects
- object model
- control points