Research map / 3D vision / NeRF and radiance fields
Neural RGB-D Surface Reconstruction
Dejan Azinović, Ricardo Martin-Brualla, Dan B Goldman, Matthias Nießner, Justus Thies
Check 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 the density-based NeRF geometry with an SDF-based surface representation and use depth supervision | NeRF-style RGB-D reconstruction · Synthetic 10 scenes | IoU: 0.550 -> 0.747 | Improved |
| Add joint pose refinement (ablation) | Neural RGB-D Surface Reconstruction · Synthetic 10 scenes | IoU: 0.655 -> 0.747 | Improved |
| Add image-plane deformation field (ablation) | Neural RGB-D Surface Reconstruction · Synthetic scene with incorrect focal length | F-score: 0.406 -> 0.904 | Improved |
| Use a 5 cm truncation width rather than 2 cm (ablation) | Neural RGB-D Surface Reconstruction · Synthetic truncation-width evaluation | C-ℓ1: 0.053 -> 0.021 | Improved |
| Use a 5 cm truncation width rather than 3 cm (ablation) | Neural RGB-D Surface Reconstruction · Synthetic truncation-width evaluation | C-ℓ1: 0.023 -> 0.021 | Improved |
| Increase truncation width from 5 cm to 10 cm (ablation) | Neural RGB-D Surface Reconstruction · Synthetic truncation-width evaluation | C-ℓ1: 0.021 -> 0.024 | Hurt |
| Add the photometric energy term to infer geometry where depth measurements are missing (ablation) | Neural RGB-D Surface Reconstruction · Synthetic scene with missing depth geometry | F-score: 0.944 -> 0.995 | Improved |
| Use color-only observations for regions without depth rather than measured depth (ablation) | Neural RGB-D Surface Reconstruction · Geometry regions with depth missing | accuracy: 8 mm -> 11 mm | Hurt |
Closest papers by meaning
- UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction (ICCV 2021 · similarity 0.91)
- NeRFPrior: Learning Neural Radiance Field as a Prior for Indoor Scene Reconstruction (CVPR 2025 · similarity 0.91)
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ECCV 2020 · similarity 0.89)
- Dense Depth Priors for Neural Radiance Fields From Sparse Input Views (CVPR 2022 · similarity 0.89)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.87)
- MorpheuS: Neural Dynamic 360deg Surface Reconstruction from Monocular RGB-D Video (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.
