Research map / 3D vision / Face and shape
C2F2NeUS: Cascade Cost Frustum Fusion for High Fidelity and Generalizable Neural Surface Reconstruction
Luoyuan Xu, Tao Guan, Yuesong Wang, Wenkai Liu, Zhaojie Zeng, Junle Wang, Wei Yang
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 per-view camera frustum volume instead of regular Euclidean volume (similar memory) (ablation) | C2F2NeUS (no cascade) · DTU | Chamfer distance: 1.77 -> 1.50 | Improved |
| increase frustum resolution from 200*144 to 400*288 (ablation) | C2F2NeUS (no cascade) · DTU | Chamfer distance: 1.50 -> 1.42 | Improved |
| increase cascade stages from 1 to 2 (ablation) | C2F2NeUS · DTU | Chamfer distance: 1.42 -> 1.28 | Improved |
| increase cascade stages from 2 to 3 (no pseudo losses) (ablation) | C2F2NeUS · DTU | Chamfer distance: 1.28 -> 1.19 | Improved |
| add pseudo-depth consistency loss L_pdc (ablation) | C2F2NeUS Stage2 · DTU | Chamfer distance: 1.28 -> 1.24 | Improved |
| add pseudo-geometry SDF loss L_pgs (ablation) | C2F2NeUS Stage2 · DTU | Chamfer distance: 1.24 -> 1.18 | Improved |
| add both pseudo losses to Stage3 (ablation) | C2F2NeUS Stage3 · DTU | Chamfer distance: 1.19 -> 1.11 | Improved |
| add cross-view and cross-level fusion instead of simple addition (ablation) | C2F2NeUS Stage3 · DTU | Chamfer distance: 2.71 -> 1.19 | Improved |
| C2F2NeUS versus generalizable SparseNeUS | C2F2NeUS · DTU 3 views | Chamfer distance: 1.64 -> 1.11 | Improved |
| increase regular Euclidean volume resolution from 96^3 to 192^3 (ablation) | Euclidean volume baseline · DTU | Chamfer distance: 1.77 -> 1.62; GPU 2231 M -> 7073 M | Mixed |
Closest papers by meaning
- Geo-Neus: Geometry-Consistent Neural Implicit Surfaces Learning for Multi-view Reconstruction (NeurIPS 2022 · similarity 0.89)
- SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration (CVPR 2024 · similarity 0.87)
- SC-NeuS: Consistent Neural Surface Reconstruction from Sparse and Noisy Views (AAAI 2024 · similarity 0.87)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.87)
- NeuSurf: On-Surface Priors for Neural Surface Reconstruction from Sparse Input Views (AAAI 2024 · similarity 0.86)
- Neural Signed Distance Function Inference through Splatting 3D Gaussians Pulled on Zero-Level Set (NeurIPS 2024 · similarity 0.86)
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.
