Research map / Point clouds / Shape and surface completion
Few ‘Zero Level Set’-Shot Learning of Shape Signed Distance Functions in Feature Space
Amine Ouasfi, Adnane Boukhayma
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 meta-learning in feature space (few-shot adaptation of decoder) | IF-Nets · ShapeNet multi-class (3000 pts, 128^3) | IoU: 0.88 -> 0.91 | Improved |
| add meta-learning in feature space | IF-Nets · ShapeNet multi-class (300 pts, 32^3) | IoU: 0.67 -> 0.74 | Improved |
| add meta-learning in feature space | IF-Nets · FAUST (trained on ShapeNet) | IoU: 0.82 -> 0.84 | Improved |
| add meta-learning (vs Ours w/o meta learn.) (ablation) | Ours · ShapeNet table class (300 pts, 32^3) | IoU: 0.58 -> 0.66 | Improved |
| add meta-learning (ablation) | Ours · ShapeNet table class (3000 pts, 128^3) | IoU: 0.82 -> 0.87 | Improved |
| add meta-learning on noisy input points (ablation) | Ours · ShapeNet table class (3000 pts, 128^3, noisy) | IoU: 0.85 -> 0.87 | Improved |
| initialize meta-decoder from pre-trained base model (ablation) | Ours · ShapeNet table class (3000 pts, 128^3) | IoU: 0.86 -> 0.87 | Improved |
| initialize meta-decoder from pre-trained base model (ablation) | Ours · ShapeNet table class (300 pts, 128^3) | CD2 (x1e-3): 0.203 -> 0.082 | Improved |
| tune the encoder during meta-learning (ablation) | Ours · ShapeNet table class (3000 pts, 128^3) | IoU: 0.87 -> 0.86 | No clear effect |
| reduce input resolution from 128^3 to 32^3 (3000 pts) | Ours vs IF-Nets · ShapeNet table class | IoU: Ours drops roughly 14% vs 20% for IF-Nets | Mixed |
| reduce input to 300 pts at 32^3 on multi-class | Ours vs IF-Nets · ShapeNet multi-class | IoU: Ours drops 18% vs 23% for IF-Nets | Mixed |
| add meta-learning inference | Ours · inference time | latency: 60 ms -> 150 ms | Hurt |
Closest papers by meaning
- MetaSDF: Meta-Learning Signed Distance Functions (NeurIPS 2020 · similarity 0.92)
- Implicit Filtering for Learning Neural Signed Distance Functions from 3D Point Clouds (ECCV 2024 · similarity 0.89)
- Few-Shot Unsupervised Implicit Neural Shape Representation Learning with Spatial Adversaries (ICML 2024 · similarity 0.89)
- LISR: Learning Linear 3D Implicit Surface Representation Using Compactly Supported Radial Basis Functions (AAAI 2024 · similarity 0.87)
- Small Steps and Level Sets: Fitting Neural Surface Models with Point Guidance (CVPR 2024 · similarity 0.86)
- Unsupervised Occupancy Learning from Sparse Point Cloud (CVPR 2024 · 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.
