ShapeLLM: Universal 3D Object Understanding for Embodied Interaction.
Zekun QiRunpei DongShaochen ZhangHaoran GengChunrui HanZheng GeHe WangLi YiKaisheng MaPublished in: CoRR (2024)
Keyphrases
- d objects
- viewpoint
- object recognition
- multi view
- pose estimation
- shape descriptors
- three dimensional
- range data
- multiple views
- cad models
- cognitive process
- object classes
- physically plausible
- line drawings
- geometric invariants
- d mesh
- object representation
- object features
- pose normalization
- image contours
- recognition and pose estimation
- topological information
- partial matching
- three dimensional objects
- control points