PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detection.
Shaoshuai ShiLi JiangJiajun DengZhe WangChaoxu GuoJianping ShiXiaogang WangHongsheng LiPublished in: CoRR (2021)
Keyphrases
- d objects
- feature set
- vector representation
- three dimensional
- feature selection
- viewpoint
- object recognition
- feature space
- classification accuracy
- feature extraction
- feature vectors
- multiple views
- pose estimation
- multi view
- shape descriptors
- minimum bounding
- fixed length
- point distribution
- similarity measure
- information retrieval
- data fusion
- document representation
- variable length
- computer vision
- data mining