Research map / Point clouds / Shape and surface completion
Score-Based Point Cloud Denoising
Shitong Luo, Wei Hu
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 score-based gradient ascent update instead of direct score addition (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 1% noise | Chamfer distance: 3.237 -> 2.521 | Improved |
| add score-based gradient ascent update instead of direct score addition (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 3% noise | Chamfer distance: 7.471 -> 4.708 | Improved |
| add iterative application to direct displacement update (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 3% noise | Chamfer distance: 7.471 -> 6.073 | Improved |
| add iterative application to direct displacement update (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 1% and 2% noise | Chamfer distance: 3.237 -> 3.237 | No clear effect |
| add neighborhood-covering training objective instead of point-position objective (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 1% noise | Chamfer distance: 4.726 -> 2.521 | Improved |
| add neighborhood-covering training objective instead of point-position objective (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 3% noise | Chamfer distance: 5.976 -> 4.708 | Improved |
| add ensemble score function instead of single score function (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 1% noise | Chamfer distance: 2.522 -> 2.521 | Improved |
| add ensemble score function instead of single score function (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 3% noise | Chamfer distance: 6.872 -> 4.708 | Improved |
| add ensemble score function instead of single score function (ablation) | Score-based point cloud denoiser · PU test set, 10K points, 3% noise | Point-to-mesh distance: 3.497 -> 1.942 | Improved |
| use denoiser trained on Gaussian noise only on unseen simulated LiDAR noise | Score-based point cloud denoiser · Simulated LiDAR noise 1% | Chamfer distance: 2.663 -> 2.616 | Improved |
| use denoising-based upsampling instead of specialized PU-Net | Score-based point cloud denoiser · PU upsampling 4x, 5K input | Chamfer distance: 3.445 -> 1.696 | Improved |
| use denoising-based upsampling instead of specialized PU-Net | Score-based point cloud denoiser · PU upsampling 4x, 10K input | Chamfer distance: 2.862 -> 1.454 | Improved |
Closest papers by meaning
- You Should Learn to Stop Denoising on Point Clouds in Advance (AAAI 2025 · similarity 0.92)
- Noise2Score3D: Tweedie's Approach for Unsupervised Point Cloud Denoising (ICCV 2025 · similarity 0.91)
- Point Cloud Upsampling Using Conditional Diffusion Module with Adaptive Noise Suppression (CVPR 2025 · similarity 0.91)
- A Conditional Denoising Diffusion Probabilistic Model for Point Cloud Upsampling (CVPR 2024 · similarity 0.90)
- U-CAN: Unsupervised Point Cloud Denoising with Consistency-Aware Noise2Noise Matching (NeurIPS 2025 · similarity 0.90)
- PointCVaR: Risk-Optimized Outlier Removal for Robust 3D Point Cloud Classification (AAAI 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.
