Research map / Point clouds / Shape and surface completion
Deep Functional Dictionaries: Learning Consistent Semantic Structures on 3D Models from Functions
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 |
|---|---|---|---|
| Use correspondence-free functional dictionaries instead of label-supervised PointNet segmentation | ShapeNet part segmentation · ShapeNet part segmentation | mean IoU: 82.4 -> 84.6 | Helped |
| Use functional dictionaries for instance proposals rather than SGPN | S3DIS instance segmentation · S3DIS, IoU threshold 0.5 | proposal recall: 64.7 -> 69.1 | Helped |
| Add the l2,1 regularizer (ablation) | ShapeNet functional dictionary · ShapeNet parts | mean IoU: 75.0 -> 84.6 | Helped |
| Increase dictionary size without the l2,1 regularizer (ablation) | ShapeNet functional dictionary · ShapeNet parts | mean IoU: 75.0; 71.2; 65.3 | Hurt |
| Omit up to fraction 0.75 of segments from training inputs (ablation) | ShapeNet functional dictionary · ShapeNet parts, partial segmentations | mean IoU: 84.6; 86.1; 86.0; 84.5 | No clear effect |
| Add input-function bit-switching noise up to probability 0.20 (ablation) | ShapeNet functional dictionary · ShapeNet parts, noisy input functions | mean IoU: 84.6; 85.8; 85.9; 85.1 | No clear effect |
| Add Siamese supervision using corresponding functions and shared coefficients (ablation) | ShapeNet functional dictionary · ShapeNet part labels matched to atoms per category | mean IoU: 77.3 -> 80.3 | Helped |
| Add Siamese supervision using corresponding functions and shared coefficients (ablation) | ShapeNet functional dictionary · ShapeNet part labels matched to atoms per object | mean IoU: 84.6 -> 85.6 | Helped |
Closest papers by meaning
- Unsupervised Learning of Intrinsic Structural Representation Points (CVPR 2020 · similarity 0.88)
- Local Deep Implicit Functions for 3D Shape (CVPR 2020 · similarity 0.88)
- Understanding and Improving Features Learned in Deep Functional Maps (CVPR 2023 · similarity 0.87)
- Learning Implicit Functions for Topology-Varying Dense 3D Shape Correspondence (NeurIPS 2020 · similarity 0.87)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.86)
- MESC-3D:Mining Effective Semantic Cues for 3D Reconstruction from a Single Image (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.