Research map / 3D vision / Human pose estimation
Disentangling Rotation and Translation from SE(3)-Equivariant Features for Shape Assembly
Hee-Jun Jung, Uigeun Ahn, JINHWI PARK, Kangil Kim
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 |
|---|---|---|---|
| replace encoder with SOT in NSM | NSM · 2BY2 two-part | RMSE-R: 37.4801 -> 29.1815 | Improved |
| replace encoder with SOT in 2BY2NET | 2BY2NET · 2BY2 two-part | CD: 2.2349 -> 1.5265 | Improved |
| SOT vs FINet-EB in NSM, part accuracy on Breaking Bad | NSM · Breaking Bad everyday | PA: 0.8824 vs 0.8887; RMSE-R 87.3287 (FINet-EB) -> 71.9768 | Mixed |
| SOT vs original encoder in SE(3)-Assembly on ShapeNet | SE(3)-Assembly · ShapeNet | PA: 0.8061 (DGCNN) -> 0.8383; RMSE-R 59.1983 -> 37.1323 | Mixed |
| SOT in dense matching CMNet | CMNet · Breaking Bad artifact | RMSE-R: 34.90 -> 34.16 | Improved |
| SOT in multi-part assembly | SE(3)-Assembly · Breaking Bad everyday multi-part | RMSE-R: 69.3440 -> 43.3528 | Improved |
| add weight regularizer Lw (ablation) | SOT · 2BY2 | RMSE-R: 34.0900 -> 32.3640 | Improved |
| add direct subtraction (ablation) | SOT · 2BY2 | CD: 0.7656 -> 0.6200 | Improved |
| add learnable projector Qt (ablation) | SOT · 2BY2 | CD: 0.7042 -> 0.6200 | Improved |
| RETI-VN instead of VNT in rotation branch (ablation) | SOT · 2BY2 | RMSE-R: 84.4165 -> 32.3640 | Improved |
| RETI-VN instead of VN in rotation branch (ablation) | SOT · 2BY2 | RMSE-R: 38.0648 -> 32.3640 | Improved |
| stack more RETI-VN layers (1 to 5) (ablation) | SOT · 2BY2 | R_inv: 0.6632 -> 0.7707 | Improved |
Closest papers by meaning
- Leveraging SE(3) Equivariance for Learning 3D Geometric Shape Assembly (ICCV 2023 · similarity 0.90)
- Shape-Pose Disentanglement Using SE(3)-Equivariant Vector Neurons (ECCV 2022 · similarity 0.89)
- Coarse-to-Fine 3D Part Assembly via Semantic Super-Parts and Symmetry-Aware Pose Estimation (NeurIPS 2025 · similarity 0.88)
- Generative 3D Part Assembly via Part-Whole-Hierarchy Message Passing (CVPR 2024 · similarity 0.88)
- Disentangling Geometric Deformation Spaces in Generative Latent Shape Models (IJCV 2023 · similarity 0.88)
- ESCAPE: Equivariant Shape Completion via Anchor Point Encoding (CVPR 2025 · similarity 0.86)
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.
