Research map / Point clouds / Shape and surface completion
Critical Regularizations for Neural Surface Reconstruction in the Wild
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 oriented MVS point-cloud supervision and Hessian and minimal-surface regularization | RegSDF · DTU | Chamfer distance (mm): VolSDF 0.857; RegSDF 0.717 | Helped |
| add oriented MVS point-cloud supervision and geometry regularizations | RegSDF · BlendedMVS | Distance F-score: SSD 71.42% -> RegSDF 69.24% | Hurt |
| add oriented MVS point-cloud supervision and geometry regularizations | RegSDF · BlendedMVS | Distance and normal F-score: SSD 61.33% -> RegSDF 59.96% | Hurt |
| add oriented MVS point-cloud supervision and geometry regularizations | RegSDF · Tanks and Temples Caterpillar | Distance and normal F-score: SIREN 15.71% -> RegSDF 16.75% | Helped |
| add oriented MVS point-cloud supervision and geometry regularizations | RegSDF · Tanks and Temples Truck | Distance and normal F-score: SIREN 37.70% -> RegSDF 42.03% | Helped |
| add Hessian regularization (ablation) | RegSDF · Tanks and Temples | Distance and normal inlier percentage: 28.51% -> 29.39% | Helped |
| add Hessian regularization (ablation) | RegSDF · DTU | Chamfer distance (mm): 0.686 -> 0.717 | No clear effect |
| add minimal surface constraint (ablation) | RegSDF · DTU | Chamfer distance (mm): 0.802 -> 0.717 | Helped |
| add minimal surface constraint (ablation) | RegSDF · Tanks and Temples | Distance and normal inlier percentage: 29.25% -> 29.39% | No clear effect |
| remove Hessian and minimal surface losses (ablation) | RegSDF · DTU | Chamfer distance (mm): 0.717 -> 0.841 | Hurt |
| remove Hessian and minimal surface losses (ablation) | RegSDF · Tanks and Temples | Distance and normal inlier percentage: 29.39% -> 23.90% | Hurt |
| increase Hessian weight to 1e-1 (ablation) | RegSDF · DTU | Chamfer distance (mm): 0.717 -> 0.939 | Hurt |
Closest papers by meaning
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.89)
- NeuSurf: On-Surface Priors for Neural Surface Reconstruction from Sparse Input Views (AAAI 2024 · similarity 0.89)
- SC-NeuS: Consistent Neural Surface Reconstruction from Sparse and Noisy Views (AAAI 2024 · similarity 0.88)
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction (CVPR 2025 · similarity 0.88)
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 2025 · similarity 0.87)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 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.