Research map / 3D vision / Face and shape
CvxNet: Learnable Convex Decomposition
Check 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 CvxNet instead of OccNet | ShapeNet/Multi Depth-to-3D reconstruction · ShapeNet/Multi Depth-to-3D | F-score: 69.08 -> 73.49 | Helped |
| Use CvxNet instead of OccNet | ShapeNet/Multi Depth-to-3D reconstruction · ShapeNet/Multi Depth-to-3D | IoU: 0.744 -> 0.731 | Hurt |
| Use CvxNet instead of SIF | ShapeNet/Multi Depth-to-3D reconstruction · ShapeNet/Multi Depth-to-3D | Chamfer-L1: 0.118 -> 0.080 | Helped |
| Use CvxNet instead of OccNet | ShapeNet/Multi RGB-to-3D reconstruction · ShapeNet/Multi RGB-to-3D | IoU: 0.571 -> 0.567 | No clear effect |
| Use CvxNet instead of AtlasNet | ShapeNet/Multi RGB-to-3D reconstruction · ShapeNet/Multi RGB-to-3D | Chamfer-L1: 0.175 -> 0.245 | Hurt |
| Use CvxNet instead of OccNet | ShapeNet/Multi RGB-to-3D reconstruction · ShapeNet/Multi RGB-to-3D | F-score: 51.75 -> 47.36 | Hurt |
| Replace indicator-function supervision with signed-distance supervision (ablation) | CvxNet · ShapeNet/Multi RGB-to-3D | reconstruction performance | Hurt |
| Merge the guidance and localization losses (ablation) | CvxNet · ShapeNet reconstruction | reconstruction performance | Hurt |
| Remove an individual loss term (ablation) | CvxNet · ShapeNet/Multi RGB-to-3D | reconstruction performance | Hurt |
| Vary the latent bottleneck dimensionality (ablation) | CvxNet · ShapeNet reconstruction | reconstruction performance | No clear effect |
| Train jointly on multiple classes instead of one class (ablation) | CvxNet Depth-to-3D · ShapeNet/Multi Depth-to-3D | reconstruction performance | Helped |
| Train on a single class instead of multiple classes (ablation) | CvxNet RGB-to-3D · ShapeNet RGB-to-3D | reconstruction performance | Helped |
Closest papers by meaning
- Learning Convex Decomposition via Feature Fields (CVPR 2026 · similarity 0.89)
- QUADify: Extracting Meshes with Pixel-level Details and Materials from Images (CVPR 2024 · similarity 0.85)
- PoNQ: a Neural QEM-based Mesh Representation (CVPR 2024 · similarity 0.85)
- Geometric Algebra Planes: Convex Implicit Neural Volumes (ICML 2025 · similarity 0.85)
- PolyhedronNet: Representation Learning for Polyhedra with Surface-attributed Graph (ICLR 2025 · similarity 0.84)
- Flatten Anything: Unsupervised Neural Surface Parameterization (NeurIPS 2024 · similarity 0.84)
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.