VelocityNet: Motion-Driven Feature Aggregation for 3D Object Detection in Point Cloud Sequences.
David EmmerichsPeter PinggeraBjörn OmmerPublished in: ICRA (2021)
Keyphrases
- d objects
- point cloud
- real world objects
- physically plausible
- cad model
- multi view
- viewpoint
- structure from motion
- object recognition
- object features
- pose estimation
- surface reconstruction
- point sets
- visual hull
- three dimensional
- multiple views
- laser scanner
- image sequences
- point distribution
- optical flow
- camera motion
- d mesh
- stereo camera
- shape descriptors
- object detection
- range data
- object classes
- feature vectors
- image features
- control points
- object surface