EFEM: Equivariant Neural Field Expectation Maximization for 3D Object Segmentation Without Scene Supervision.
Jiahui LeiCongyue DengKarl SchmeckpeperLeonidas J. GuibasKostas DaniilidisPublished in: CoRR (2023)
Keyphrases
- d objects
- cad models
- expectation maximization
- three dimensional
- object recognition
- real world objects
- image segmentation
- multi view
- object contours
- pose estimation
- viewpoint
- surface patches
- object parts
- range data
- em algorithm
- object features
- level set
- multiple objects
- geometric models
- partial matching
- segmentation algorithm
- multiple views
- viewing direction
- multiple images
- cad model
- shape descriptors
- intensity images
- image contours
- segmentation method
- surface properties
- point distribution
- object classes
- d scene
- single image
- video sequences
- generative model
- d mesh
- physically plausible
- moving objects
- multi views
- image sequences
- topological information
- maximum likelihood
- line drawings
- medical images
- active contours
- integral imaging
- pose normalization
- structured light
- object models