Research map / Point clouds / Shape and surface completion
Temporally-Consistent Surface Reconstruction Using Metrically-Consistent Atlases
Jan Bednařík, Noam Aigerman, Vladimir G. Kim, Siddhartha Chaudhuri, Shaifali Parashar, Mathieu Salzmann, Pascal Fua
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 |
|---|---|---|---|
| use adjacent instead of random training pairs (ablation) | OUR · AMA (crane) | m_sL2: 226.21 -> 46.92 | Improved |
| use adjacent instead of random training pairs (ablation) | OUR · ANIM (cat) | m_sL2: 54.48 -> 8.45 | Improved |
| use adjacent instead of random training pairs (ablation) | OUR · DFAUST (jacks) | m_sL2: 25.06 -> 23.96 | Improved |
| add L_metric (alpha_mc=0.1 versus 1e-3) (ablation) | OUR · AMA (crane) | m_sL2: 56.62 -> 46.92 | Improved |
| too large alpha_mc (10 versus 0.1) (ablation) | OUR · AMA (crane) | m_sL2: 46.92 -> 253.14 | Hurt |
| too large alpha_rot (10 versus 0.1) (ablation) | OUR · AMA (crane) | m_sL2: 46.92 -> 181.73 | Hurt |
| add L_metric to the vanilla model (ablation) | OUR · AMA (crane) | m_sL2: 317.02 -> 247.44 | Improved |
| add L_rot to the vanilla model (ablation) | OUR · AMA (crane) | m_sL2: 317.02 -> 95.47 | Improved |
| add both L_metric and L_rot (ablation) | OUR · AMA (crane) | m_sL2: 317.02 -> 46.92 | Improved |
| add progressive sequence sampling (ablation) | OUR · AMA (crane) | m_sL2: 268.97 -> 46.92 | Improved |
| add L_rot with progressive sampling on clean aligned data (ablation) | OUR · ANIM | m_sL2: 11.93 -> 12.60 | No clear effect |
| add L_rot with progressive sampling on clean aligned data (ablation) | OUR · AMAa | m_sL2: 57.12 -> 35.60 | Improved |
Closest papers by meaning
- Temporally-Coherent Surface Reconstruction via Metric-Consistent Atlases (ICCV 2021 · similarity 0.97)
- DynoSurf: Neural Deformation-based Temporally Consistent Dynamic Surface Reconstruction (ECCV 2024 · similarity 0.88)
- Deep Geometric Prior for Surface Reconstruction (CVPR 2019 · similarity 0.88)
- Surface Reconstruction From Point Clouds: A Survey and a Benchmark (TPAMI 2024 · similarity 0.87)
- OffsetOPT: Explicit Surface Reconstruction without Normals (CVPR 2025 · similarity 0.85)
- Occlusion-aware Non-Rigid Point Cloud Registration via Unsupervised Neural Deformation Correntropy (ICLR 2025 · similarity 0.85)
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.
