Research map / 3D vision / Face and shape
Voxurf: Voxel-based Efficient and Accurate Neural Surface Reconstruction
Read the paperCheck 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 dual color network (ablation) | SDF voxel baseline · DTU (15 scans, masked) | Chamfer Distance mean: 0.91 -> 0.79 | Helped |
| add hierarchical geometry feature (ablation) | SDF voxel baseline · DTU (15 scans, masked) | Chamfer Distance mean: 0.91 -> 0.77 | Helped |
| add dual color network and hierarchical geometry feature together (ablation) | SDF voxel baseline · DTU (15 scans, masked) | Chamfer Distance mean: 0.91 -> 0.72 | Helped |
| add residual and detach design to dual color network (ablation) | dual color network · DTU (15 scans, masked) | Chamfer Distance mean: 0.77 -> 0.72 (residual+detach vs. neither) | Helped |
| add SDF TV and gradient smoothness loss in fine stage (ablation) | Voxurf fine stage · DTU (15 scans, masked) | Chamfer Distance mean: 1.18 -> 0.72 (with both, vs none) | Helped |
| apply Gaussian kernel for post-processing of SDF before extraction (ablation) | Voxurf · DTU (15 scans, masked) | Chamfer Distance mean: 0.720 -> 0.715 when removed | No clear effect |
| apply Gaussian kernel during coarse training (ablation) | Voxurf · DTU (15 scans, masked) | Chamfer Distance mean: 0.72 -> 0.74 when removed | Helped |
| increase fine-stage voxel resolution from 192^3 to 256^3 (ablation) | Voxurf · DTU (15 scans, masked) | Chamfer Distance mean: 0.75 -> 0.72 | Helped |
| increase fine-stage voxel resolution from 256^3 to 320^3 (ablation) | Voxurf · DTU (15 scans, masked) | Chamfer Distance mean: 0.72 -> 0.73 | No clear effect |
| add dual color network and hierarchical feature to DVGO+NeuS baseline with smoothness priors | DVGO + NeuS · DTU (15 scans, masked) | Chamfer Distance mean: 1.13 -> 0.72 (Voxurf vs naive DVGO+NeuS) | Helped |
| Voxurf vs NeuS on DTU masked surface reconstruction | Voxurf · DTU (15 scans, masked) | Chamfer Distance mean: 0.77 -> 0.72 | Helped |
Closest papers by meaning
- GeoSVR: Taming Sparse Voxels for Geometrically Accurate Surface Reconstruction (NeurIPS 2025 · similarity 0.91)
- NeAT: Learning Neural Implicit Surfaces With Arbitrary Topologies From Multi-View Images (CVPR 2023 · similarity 0.91)
- HF-NeuS: Improved Surface Reconstruction Using High-Frequency Details (NeurIPS 2022 · similarity 0.91)
- DySurface: Consistent 4D Surface Reconstruction via Bridging Explicit Gaussians and Implicit Functions (NeurIPS 2026 · similarity 0.89)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.89)
- Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning (AAAI 2024 · similarity 0.89)
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.