Research map / Point clouds / Shape and surface completion
Reconstructing Surfaces for Sparse Point Clouds With On-Surface Priors
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 an unsigned-distance on-surface prior and geometric regularization | SDF reconstruction from sparse point clouds · ShapeNet (8 classes, mean) | L1CD×10: 0.106 -> 0.068 | Helped |
| use a UDF rather than an SDF as the on-surface prior (ablation) | SDF reconstruction on ABC · ABC (10 test shapes, 500 points) | L1CD: 0.049 -> 0.015 | Helped |
| use a UDF rather than a binary classifier as the on-surface prior (ablation) | SDF reconstruction on ABC · ABC (10 test shapes, 500 points) | L1CD: 0.083 -> 0.015 | Helped |
| add geometric regularization (λ=0.4) instead of omitting it (ablation) | SDF reconstruction on ABC · ABC (10 test shapes, 500 points) | L1CD: 0.050 -> 0.015 | Helped |
| increase the geometric-regularization weight from 0.2 to 0.4 (ablation) | SDF reconstruction on ABC · ABC (10 test shapes, 500 points) | L1CD: 0.016 -> 0.015 | Helped |
| increase the geometric-regularization weight from 0.4 to 0.6 (ablation) | SDF reconstruction on ABC · ABC (10 test shapes, 500 points) | L1CD: 0.015 -> 0.023 | Hurt |
| use 50 neighbors rather than 25, 100, or 200 (ablation) | On-surface prior · ABC (10 test shapes, 500 points) | L1CD: 0.016 -> 0.015; 0.016 -> 0.015; 0.018 -> 0.015 | Helped |
| add translation normalization (ablation) | On-surface prior · ABC (10 test shapes, 500 points) | L1CD: 0.084 -> 0.015 | Helped |
| add rotation alignment to translation normalization (ablation) | On-surface prior · ABC (10 test shapes, 500 points) | L1CD: 0.015 -> 0.035 | Hurt |
| replace the neighbor-feature MLP with PointNet (ablation) | On-surface prior · ABC (10 test shapes, 500 points) | L1CD: 0.015 -> 0.076 | Hurt |
| replace the neighbor-feature MLP with PointNet++ (ablation) | On-surface prior · ABC (10 test shapes, 500 points) | L1CD: 0.015 -> 0.070 | Hurt |
| apply the prior trained at density 500 to denser inputs (ablation) | On-surface prior · ABC test shapes, density 500 to 2000 | L1CD: 0.015 -> 0.012 | Helped |
Closest papers by meaning
- Learning Bijective Surface Parameterization for Inferring Signed Distance Functions from Sparse Point Clouds with Grid Deformation (CVPR 2025 · similarity 0.91)
- Inferring Neural Signed Distance Functions by Overfitting on Single Noisy Point Clouds through Finetuning Data-Driven based Priors (NeurIPS 2024 · similarity 0.90)
- MultiPull: Detailing Signed Distance Functions by Pulling Multi-Level Queries at Multi-Step (NeurIPS 2024 · similarity 0.90)
- A Lightweight UDF Learning Framework for 3D Reconstruction Based on Local Shape Functions (CVPR 2025 · similarity 0.89)
- OffsetOPT: Explicit Surface Reconstruction without Normals (CVPR 2025 · similarity 0.88)
- Toward Robust Neural Reconstruction from Sparse Point Sets (CVPR 2025 · similarity 0.88)
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.