Research map / Point clouds / Shape and surface completion
PrimitiveUDF: Primitive-Based Unsigned Distance Fields for Surface Reconstruction from Point Clouds
Xin Deng, Bo Yang, Yifei Shi, Bing Wang
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 |
|---|---|---|---|
| add spherical reorganization to primitive construction (ablation) | PrimitiveUDF · ScanNet | CDL1 mean (x1e-3): 2.11 -> 1.86 | Improved |
| add primitive construction on top of spherical reorganization-based neighborhood (ablation) | PrimitiveUDF · ScanNet | CDL1 mean (x1e-3): 1.91 (Base) -> 1.86 | Improved |
| add primitive construction without spherical reorganization (ablation) | Base UDF regressor · ScanNet | CDL1 mean (x1e-3): 1.91 -> 2.11 | Hurt |
| use triangle primitives instead of segment primitives (ablation) | PrimitiveUDF · ScanNet | CDL2 mean (x1e-6): 6.92 -> 6.03 | Improved |
| use triangle instead of four-point patch primitives (ablation) | PrimitiveUDF · ScanNet | CDL2 mean (x1e-6): 7.54 -> 6.03 | Improved |
| add segment primitives to triangle primitives (ablation) | PrimitiveUDF · ScanNet | CDL1 mean (x1e-3): 1.86 -> 1.86 | No clear effect |
| PrimitiveUDF vs GeoUDF on watertight ShapeNet | PrimitiveUDF · Watertight ShapeNet (clean) | F-score@0.5%: 93.8 -> 95.0 | Improved |
| PrimitiveUDF vs GeoUDF on watertight ShapeNet noisy | PrimitiveUDF · Watertight ShapeNet (noise 0.005) | F-score@0.5%: 89.3 -> 91.7 | Improved |
| PrimitiveUDF vs best baseline on ShapeNet Cars | PrimitiveUDF · ShapeNet Cars (MC 256) | CDL2 mean (x1e-4): 0.120 -> 0.110 | Improved |
| PrimitiveUDF vs SALS on ABC | PrimitiveUDF · ABC | CDL1 (x1e-2): 0.251 -> 0.251 | No clear effect |
| PrimitiveUDF vs GeoUDF on ScanNet | PrimitiveUDF · ScanNet (MC 256) | CDL1 mean (x1e-3): 2.05 -> 1.86 | Improved |
| PrimitiveUDF vs SALS at resolution 128 on ScanNet | PrimitiveUDF · ScanNet (MC 128) | F-score@0.5%: 94.7 (SALS) vs 94.3 | Hurt |
Closest papers by meaning
- Learning Consistency-Aware Unsigned Distance Functions Progressively from Raw Point Clouds (NeurIPS 2022 · similarity 0.90)
- GeoUDF: Surface Reconstruction from 3D Point Clouds via Geometry-guided Distance Representation (ICCV 2023 · similarity 0.90)
- A Lightweight UDF Learning Framework for 3D Reconstruction Based on Local Shape Functions (CVPR 2025 · similarity 0.90)
- CAP-UDF: Learning Unsigned Distance Functions Progressively From Raw Point Clouds With Consistency-Aware Field Optimization (TPAMI 2024 · similarity 0.90)
- GaussianUDF: Inferring Unsigned Distance Functions through 3D Gaussian Splatting (CVPR 2025 · similarity 0.88)
- Details Enhancement in Unsigned Distance Field Learning for High-fidelity 3D Surface Reconstruction (AAAI 2025 · 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.
