Deep Octree-based CNNs with Output-Guided Skip Connections for 3D Shape and Scene Completion.
Peng-Shuai WangYang LiuXin TongPublished in: CVPR Workshops (2020)
Keyphrases
- single image
- d objects
- three dimensional
- d scene
- video sequences
- multiple images
- geometric primitives
- cellular neural networks
- dynamic scenes
- surface geometry
- shape reconstruction
- wire frame
- image sequences
- hierarchical representation
- complex scenes
- data structure
- scene understanding
- visual hull
- multi view
- moving objects
- synthesized images
- geometric constraints
- real scenes
- super resolution
- viewpoint
- object recognition
- rigid body motion