Research map / Point clouds / Shape and surface completion
VKSR: Scalable Kernel Surface Reconstruction Using Vecchia's Approximation
Maximilian Weiherer, Chukwudi Williams Umah, Bernhard Egger
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 |
|---|---|---|---|
| replace low-rank Nystrom approximation with Vecchia local approximation | kernel surface reconstruction · ShapeNet dense (100K points) | runtime (s): 47.91 -> 0.77 | Improved |
| replace low-rank Nystrom approximation with Vecchia local approximation | Matern kernel method · ShapeNet dense (100K points) | Chamfer distance (x10^3): 2.25 -> 2.19 | Improved |
| replace low-rank approximation with Vecchia on sparse inputs | kernel methods · ShapeNet sparse (1K points) | Chamfer distance (x10^3): 3.90 -> 3.95 | Hurt |
| use VKSR instead of SPSR on sparse inputs | SPSR · ShapeNet sparse (1K points) | Chamfer distance (x10^3): 4.36 -> 3.95 | Improved |
| use VKSR instead of SPSR on dense inputs | SPSR · ShapeNet dense (100K points) | Chamfer distance (x10^3): 2.04 -> 2.19 | Hurt |
| use VKSR instead of NS on noisy incomplete scans | NS · Surface Reconstruction Benchmark (Anchor) | runtime (s): 308.86 -> 13.79 | Improved |
| use VKSR instead of SPSR on noisy incomplete scans | SPSR · Surface Reconstruction Benchmark (Anchor) | Chamfer distance: 0.30 -> 0.24 | Improved |
| use approximate nearest neighbor search (ablation) | VKSR · Stanford 3D Scanning Repository | runtime: Lucy 1809.79 -> 80.18 s | Improved |
| increase number of neighbors k (ablation) | VKSR · ShapeNet | Chamfer distance (x10^3): 2.25 -> 2.15 | Improved |
| use VKSR instead of SPSR on ScanNet scenes | SPSR · ScanNet (20 scenes) | Chamfer distance: 0.02 -> 0.02 | No clear effect |
| use VKSR instead of chunked subdivision of NS | NS · Stanford Lucy (14M points) | Chamfer distance: 3.13 -> 0.81 | Improved |
Closest papers by meaning
- Neural Kernel Surface Reconstruction (CVPR 2023 · similarity 0.91)
- Matérn Kernels for Tunable Implicit Surface Reconstruction (ICLR 2025 · similarity 0.87)
- A Linear Time and Space Local Point Cloud Geometry Encoder via Vectorized Kernel Mixture (VecKM) (ICML 2024 · similarity 0.87)
- SPAR3D: Stable Point-Aware Reconstruction of 3D Objects from Single Images (CVPR 2025 · similarity 0.86)
- GVKF: Gaussian Voxel Kernel Functions for Highly Efficient Surface Reconstruction in Open Scenes (NeurIPS 2024 · similarity 0.86)
- GeoSVR: Taming Sparse Voxels for Geometrically Accurate Surface Reconstruction (NeurIPS 2025 · 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.
