Research map / Image generation / Image restoration
Learning Self-Prior for Mesh Denoising Using Dual Graph Convolutional Networks
Shota Hattori, Tatsuya Yatagawa, Yutaka Ohtake, Hiromasa Suzuki
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 |
|---|---|---|---|
| train PosNet and NormNet jointly with consistency loss (full DDMP instead of PosNet alone) (ablation) | DDMP · grayloc | AAD: 9.19 -> 7.25 | Improved |
| add PosNet to NormNet (full DDMP vs NormNet alone) (ablation) | DDMP · grayloc | AAD: 10.50 -> 7.25 | Improved |
| add position-normal consistency loss Econ (ablation) | DDMP · grayloc | AAD: 8.89 -> 7.25 | Improved |
| add position-normal consistency loss Econ (ablation) | DDMP · carter | AAD: 5.09 -> 4.95 | Improved |
| add Laplacian error ELap to PosNet (ablation) | PosNet · grayloc | AAD: 26.22 -> 9.19 | Improved |
| add Laplacian error ELap to PosNet (ablation) | PosNet · carter | AAD: 9.27 -> 4.96 | Improved |
| add bilateral smoothness error Ebnf to NormNet (ablation) | NormNet · grayloc | AAD: 14.05 -> 10.50 | Improved |
| add bilateral smoothness error Ebnf to NormNet (ablation) | NormNet · carter | AAD: 4.88 -> 4.89 | No clear effect |
| use 30-step Laplacian smoothing without cotangent weights instead of no smoothing (ablation) | DDMP · grayloc | AAD: 34.98 -> 7.25 | Improved |
| use LS-no-CW instead of BNF initial smoothing (ablation) | DDMP · grayloc (non-CAD) | AAD: 9.79 -> 7.25 | Improved |
| use BNF instead of LS-no-CW initial smoothing (ablation) | DDMP · sharp (CAD) | AAD: 4.90 -> 4.64 | Improved |
| use DDMP self-prior instead of pretrained GCN-D on CAD mesh block | mesh denoising · block | AAD: 4.19 -> 2.90 | Improved |
Closest papers by meaning
- Universal Denoising Networks : A Novel CNN Architecture for Image Denoising (CVPR 2018 · similarity 0.89)
- Hyper-MD: Mesh Denoising with Customized Parameters Aware of Noise Intensity and Geometric Characteristics (CVPR 2024 · similarity 0.89)
- Self-Supervised Image Denoising Using Implicit Deep Denoiser Prior (AAAI 2023 · similarity 0.89)
- Meshlet Priors for 3D Mesh Reconstruction (CVPR 2020 · similarity 0.89)
- Local Surface Descriptor for Geometry and Feature Preserved Mesh Denoising (AAAI 2022 · similarity 0.88)
- Primal-Dual Mesh Convolutional Neural Networks (NeurIPS 2020 · 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.
