Research map / Point clouds / Point cloud classification
Canonical Capsules: Self-Supervised Capsules in Canonical Pose
Weiwei Sun, Andrea Tagliasacchi, Boyang Deng, Sara Sabour, Soroosh Yazdani, Geoffrey Hinton, Kwang Moo Yi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add equivariance loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.16 -> 1.11 | Improved |
| add equivariance loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | mStd: 110.174 -> 8.278 | Improved |
| add localization loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.44 -> 1.11 | Improved |
| add equilibrium loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.60 -> 1.11 | Improved |
| add invariance loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.12 -> 1.11 | No clear effect |
| add canonical loss (ablation) | Canonical Capsules · ShapeNet airplane unaligned | mStd: 8.421 -> 8.278 | Improved |
| use ACNe backbone instead of PointNet (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.28 -> 1.11 | Improved |
| use 10 capsules instead of 5 (constant capacity) (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.25 -> 1.11 | Improved |
| use 10 capsules instead of 20 (constant capacity) (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.15 -> 1.11 | Improved |
| use learned canonicalizer instead of one-shot alignment (ablation) | Canonical Capsules · ShapeNet chair unaligned | Chamfer distance: 2.92 -> 2.58 | Improved |
| use learned canonicalizer instead of one-shot alignment (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 1.10 -> 1.11 | No clear effect |
| re-extract descriptors after canonicalization (ablation) | Canonical Capsules · ShapeNet airplane unaligned | Chamfer distance: 2.12 -> 1.08 | Improved |
Later papers that compared with this method
1 later accepted paper reports Canonical Capsules as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Shape-Pose Disentanglement Using SE(3)-Equivariant Vector Neurons (ECCV 2022 · better on 3 of 4 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DeepGMR — DeepGMR: Learning Latent Gaussian Mixture Models for Registration (ECCV 2020) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ConDor: Self-Supervised Canonicalization of 3D Pose for Partial Shapes (CVPR 2022 · similarity 0.90)
- EquiCaps: Predictor-Free Pose-Aware Pre-Trained Capsule Networks (ICCV 2025 · similarity 0.88)
- 3D Point Capsule Networks (CVPR 2019 · similarity 0.88)
- Learning to Orient Surfaces by Self-supervised Spherical CNNs (NeurIPS 2020 · similarity 0.88)
- SIEVE: Overcoming Topological Obstruction in Equivariant Self-Supervised Learning (NeurIPS 2026 · similarity 0.87)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
