Research map / 3D vision / Face and shape
Learning Local Neighboring Structure for Robust 3D Shape Representation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace COMA's convolution with LSA-Conv | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 5.238 -> 3.492 | Helped |
| Replace Spiral's convolution with LSA-Conv | LSA-3DMM · COMA test | L2 reconstruction error (mm): 0.227 -> 0.117 | Helped |
| Use an LSA-Conv decoder instead of the prior decoder | 3D shape correspondence system · FAUST-inter | Average Euclidean distance (cm): 2.878cm -> 2.501cm | Helped |
| Factorize weighting matrices using a subspace with B=8 (ablation) | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 5.110 -> 4.544; 4.667 -> 4.544 | Helped |
| Factorize weighting matrices using a subspace with B=8 (ablation) | LSA-3DMM · COMA test | L2 reconstruction error (mm): 0.198 -> 0.179; 0.193 -> 0.179 | Helped |
| Randomly reshuffle each node's neighbor order (ablation) | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 3.492 -> 3.527 | No clear effect |
| Randomly reshuffle each node's neighbor order (ablation) | LSA-3DMM · COMA test | L2 reconstruction error (mm): 0.117 -> 0.117 | No clear effect |
| Initialize weighting matrices randomly instead of with identity matrices (ablation) | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 3.492 -> 4.488 | Hurt |
| Initialize weighting matrices randomly instead of with identity matrices (ablation) | LSA-3DMM · COMA test | L2 reconstruction error (mm): 0.117 -> 0.137 | Hurt |
| Add learnable weighting matrices for each node's neighbors (ablation) | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 5.511 -> 3.492 | Helped |
| Add learnable weighting matrices for each node's neighbors (ablation) | LSA-3DMM · COMA test | L2 reconstruction error (mm): 0.212 -> 0.117 | Helped |
| Reduce neighbor size to K=5 (ablation) | LSA-3DMM · DFAUST test | L2 reconstruction error (mm): 3.492 -> 3.869 | Hurt |
Closest papers by meaning
- FeaStNet: Feature-Steered Graph Convolutions for 3D Shape Analysis (CVPR 2018 · similarity 0.89)
- GEOMetrics: Exploiting Geometric Structure for Graph-Encoded Objects (ICML 2019 · similarity 0.88)
- Local Implicit Grid Representations for 3D Scenes (CVPR 2020 · similarity 0.87)
- Neural Parts: Learning Expressive 3D Shape Abstractions With Invertible Neural Networks (CVPR 2021 · similarity 0.87)
- Im2Struct: Recovering 3D Shape Structure From a Single RGB Image (CVPR 2018 · similarity 0.87)
- Common3D: Self-Supervised Learning of 3D Morphable Models for Common Objects in Neural Feature Space (CVPR 2025 · similarity 0.86)
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.