Research map / 3D vision / Face and shape
PatchRD: Detail-Preserving Shape Completion by Learning Patch Retrieval and Deformation
Bo Sun, Vladimir G. Kim, Noam Aigerman, Qixing Huang, Siddhartha Chaudhuri
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 learned deformation instead of ICP patch alignment (ablation) | PatchRD · Random-crop ShapeNet, chair | CD: 4.90 -> 1.43 | Improved |
| add learned deformation instead of ICP patch alignment (ablation) | PatchRD · Random-crop ShapeNet, chair | FID: 43.35 -> 11.89 | Improved |
| blend several retrieved patches instead of placing best one (ablation) | PatchRD · Random-crop ShapeNet, chair | CD: 2.03 -> 1.43 | Improved |
| blend several retrieved patches instead of placing best one (ablation) | PatchRD · Random-crop ShapeNet, chair | FID: 30.25 -> 11.89 | Improved |
| add blending smoothness loss (ablation) | PatchRD · Random-crop ShapeNet, chair | CD: 1.86 -> 1.43 | Improved |
| add blending smoothness loss (ablation) | PatchRD · Random-crop ShapeNet, chair | FID: 27.42 -> 11.89 | Improved |
| use PatchRD instead of SnowflakeNet | SnowflakeNet · Random-crop ShapeNet, average | CD: 2.06 -> 1.22 | Improved |
| use PatchRD instead of SnowflakeNet on planes | SnowflakeNet · Random-crop ShapeNet, Plane | CD: 0.72 -> 0.98 | Hurt |
| use PatchRD instead of 3D-GAN | 3D-GAN · Random-crop ShapeNet, chair | FID: 157.19 -> 11.89 | Improved |
| use PatchRD instead of Conv-ONet | Conv-ONet · Random-crop ShapeNet, chair | FID: 174.72 -> 11.89 | Improved |
| increase crop size from 10% to 60% (ablation) | PatchRD · Random-crop ShapeNet | CD: 0.88 -> 6.64 | Hurt |
| increase crop size from 10% to 60% (ablation) | PatchRD · Random-crop ShapeNet | FID: 9.74 -> 15.63 | Hurt |
Closest papers by meaning
- PatchComplete: Learning Multi-Resolution Patch Priors for 3D Shape Completion on Unseen Categories (NeurIPS 2022 · similarity 0.92)
- PatchNets: Patch-Based Generalizable Deep Implicit 3D Shape Representations (ECCV 2020 · similarity 0.90)
- DiffComplete: Diffusion-based Generative 3D Shape Completion (NeurIPS 2023 · similarity 0.89)
- ShapeFormer: Transformer-Based Shape Completion via Sparse Representation (CVPR 2022 · similarity 0.88)
- Shape Completion Using 3D-Encoder-Predictor CNNs and Shape Synthesis (CVPR 2017 · similarity 0.88)
- DinoComplete: 3D Shape Completion with Distilled Semantic Priors and State Space Models (NeurIPS 2026 · 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.
