3D Point Capsule Networks
About
In this paper, we propose 3D point-capsule networks, an auto-encoder designed to process sparse 3D point clouds while preserving spatial arrangements of the input data. 3D capsule networks arise as a direct consequence of our novel unified 3D auto-encoder formulation. Their dynamic routing scheme and the peculiar 2D latent space deployed by our approach bring in improvements for several common point cloud-related tasks, such as object classification, object reconstruction and part segmentation as substantiated by our extensive evaluations. Moreover, it enables new applications such as part interpolation and replacement.
Yongheng Zhao, Tolga Birdal, Haowen Deng, Federico Tombari• 2018
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| 3D Object Classification | ModelNet40 (test) | Accuracy93.2 | 302 | |
| 3D Point Cloud Classification | ModelNet40 (test) | OA88.9 | 297 | |
| 3D Shape Classification | ModelNet40 (test) | Accuracy89.3 | 227 | |
| Object Classification | ModelNet40 (test) | -- | 180 | |
| 3D Object Part Segmentation | ShapeNet Part (test) | -- | 114 | |
| Classification | ModelNet40 (test) | Accuracy88.9 | 99 | |
| Few-shot 3D Object Classification (5-way) | ModelNet40 (test) | 10-shot Accuracy42.3 | 57 | |
| Few-shot 3D Object Classification (10-way) | ModelNet40 (test) | Accuracy (10-shot)38 | 46 | |
| Few-shot object classification | ModelNet40 (test) | Accuracy (5-way, 10-shot)42.3 | 29 | |
| Descriptor matching | 3DMatch Rotated | -- | 18 |
Showing 10 of 16 rows