Research map / Point clouds / Shape and surface completion
Aligning Gradient and Hessian for Neural Signed Distance Function
Ruian Wang, Zixiong Wang, Yunxiao Zhang, Shuangmin Chen, Shiqing Xin, Changhe Tu, Wenping 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 gradient-Hessian alignment loss | SIREN · SRB | Chamfer: 18.24 -> 3.86 | Improved |
| add gradient-Hessian alignment loss | SIREN · ABC | F-Score: 83.50 vs 30.95 | Improved |
| add gradient-Hessian alignment loss | SIREN · 3D Scene | F-Score: 53.94 -> 58.37 | Improved |
| add gradient-Hessian alignment loss | Neural-Pull · 3D Scene | Chamfer: 33.96 -> 12.04 | Improved |
| add gradient-Hessian alignment loss | SIREN · 3D Scene | Chamfer: 7.38 -> 7.14 | Improved |
| compare unsupervised method against supervised POCO | Ours (SIREN) · ShapeNet 3K points | IOU: 94.40 vs 89.80 | Hurt |
| alignment loss weight alpha 0.06 -> 6 (ablation) | Ours (SIREN) · ShapeNet | F-Score: 76.82 -> 88.71 | Improved |
| alignment loss weight alpha 6 -> 60 (ablation) | Ours (SIREN) · ShapeNet | F-Score: 88.71 -> 82.91 | Hurt |
| alignment loss weight alpha 6 -> 60 (ablation) | Ours (SIREN) · ShapeNet | Chamfer: 3.38 -> 2.19 | Improved |
| adaptive weight decay delta 0 -> 10 (ablation) | Ours (SIREN) · ShapeNet | F-Score: 79.95 -> 88.71 | Improved |
| adaptive weight decay delta 10 -> 100 (ablation) | Ours (SIREN) · ShapeNet | F-Score: 88.71 -> 80.12 | Hurt |
| replace alignment loss with Dirichlet energy (ablation) | Ours (SIREN) · ShapeNet | F-Score: 88.71 -> 77.67 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DiGS — DiGS: Divergence Guided Shape Implicit Neural Representation for Unoriented Point Clouds (CVPR 2022) · better on 9 of 9 shared results
- POCO+ — POCO: Point Convolution for Surface Reconstruction (CVPR 2022) · better on 3 of 4 shared results
- Neural-Pull — Neural-Pull: Learning Signed Distance Function from Point clouds by Learning to Pull Space onto Surface (ICML 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Small Steps and Level Sets: Fitting Neural Surface Models with Point Guidance (CVPR 2024 · similarity 0.90)
- Learning Bijective Surface Parameterization for Inferring Signed Distance Functions from Sparse Point Clouds with Grid Deformation (CVPR 2025 · similarity 0.90)
- Inferring Neural Signed Distance Functions by Overfitting on Single Noisy Point Clouds through Finetuning Data-Driven based Priors (NeurIPS 2024 · similarity 0.89)
- Toward Robust Neural Reconstruction from Sparse Point Sets (CVPR 2025 · similarity 0.88)
- Sharpening Neural Implicit Functions with Frequency Consolidation Priors (AAAI 2025 · similarity 0.88)
- Few-Shot Unsupervised Implicit Neural Shape Representation Learning with Spatial Adversaries (ICML 2024 · 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.
