Representation:
- N × 3 N\times3 N×3 matrix or N × ( 3 + D ) N\times(3+D) N×(3+D) matrix
Data sources:
- Lidar
- RGB-D
- CAD Models
- SfM
- Depth from Images
Applications:
- Localization: SLAM, loop closure, registration
- Perception: object detection, classification
- Reconstruction: SfM, registration
Options:
- Mesh
- Voxel Grid
- Octree
- Point Cloud
Difficulties:
- Sparsity
- Irregular: difficulty in neighbor searching
- Lack of texture information
- Un-ordered: difficulty in deep learning
- Rotation equivariance / invariance