Research map / 3D vision / Human pose estimation
Shape Prior Deformation for Categorical 6D Object Pose and Size Estimation
Meng Tian, Marcelo H Ang Jr, Gim Hee Lee
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 |
|---|---|---|---|
| regress NOCS coordinates directly instead of shape-deformation plus correspondence (ablation) | Ours (RGB-D) · CAMERA25 | mAP 5deg 2cm: -3.1% | Improved |
| regress NOCS coordinates directly instead of shape-deformation plus correspondence (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: -5.6% | Improved |
| add deformation regularization (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 19.3 -> 13.4 | Improved |
| add entropy regularization (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 19.3 -> 15.7 | Improved |
| add regularization losses (ablation) | Ours (RGB-D) · CAMERA25 | mAP 5deg 2cm: 54.3 -> 53.9 | No clear effect |
| use learned categorical shape prior instead of sphere (no prior) (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 15.6 -> 19.3 | Improved |
| use learned categorical shape prior instead of sphere (no prior) (ablation) | Ours (RGB-D) · CAMERA25 | mAP 5deg 2cm: 54.0 -> 54.3 | No clear effect |
| use mean latent embedding prior instead of random instance prior (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 17.9 -> 19.3 | Improved |
| use mean latent embedding prior instead of NN instance prior (ablation) | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 17.0 -> 19.3 | Improved |
| add depth input to network | Ours (RGB-D) · CAMERA25 | mAP 5deg 2cm: 50.2 -> 54.3 | Improved |
| add depth input to network | Ours (RGB-D) · REAL275 | mAP 5deg 2cm: 15.7 -> 19.3 | Improved |
| deform shape prior instead of using it as is for reconstruction | Reconstruction network · CAMERA25 shape reconstruction | Chamfer distance x1e-3: 3.70 -> 1.97 | Improved |
Closest papers by meaning
- Category-Level 6D Object Pose and Size Estimation Using Self-Supervised Deep Prior Deformation Networks (ECCV 2022 · similarity 0.92)
- Learning Canonical Shape Space for Category-Level 6D Object Pose and Size Estimation (CVPR 2020 · similarity 0.91)
- FS-Net: Fast Shape-Based Network for Category-Level 6D Object Pose Estimation With Decoupled Rotation Mechanism (CVPR 2021 · similarity 0.90)
- PriorPose: Reference-Guided Joint Deformation and Alignment for Category-Level Object Pose Estimation (ECCV 2026 · similarity 0.89)
- Learning Category-Specific Deformable 3D Models for Object Reconstruction (TPAMI 2016 · similarity 0.89)
- CatFormer: Category-Level 6D Object Pose Estimation with Transformer (AAAI 2024 · 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.
