Research map / 3D vision / Face and shape
Learning Descriptor Networks for 3D Shape Synthesis and Analysis
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 3D DescriptorNet instead of 3D VAE for synthesis | ModelNet10 synthesis across 10 categories · ModelNet10, 10 categories | Inception score: 11.015$\pm$0.420 -> 11.772$\pm$0.418 | Helped |
| Use 3D DescriptorNet instead of the strongest reported baseline | Single-category shape synthesis on ModelNet10 · ModelNet10, single-category synthesis (10 categories) | Average softmax class probability: 0.7431 -> 0.8811 | Helped |
| Use conditional 3D DescriptorNet instead of the strongest reported baseline | 3D object recovery · ModelNet10, 70% occlusion | Recovery error: 0.0259 -> 0.0092 | Helped |
| Use unsupervised 3D DescriptorNet features instead of PANORAMA-NN | ModelNet10 classification · ModelNet10 | Accuracy: 91.1\% -> 92.4\% | Helped |
| Disable the Langevin noise term in the later training stage (ablation) | 3D DescriptorNet synthesis · ModelNet10 shape synthesis | Synthesis quality | Helped |
Closest papers by meaning
- Unsupervised Deep Shape Descriptor With Point Distribution Learning (CVPR 2020 · similarity 0.88)
- Shape Completion Using 3D-Encoder-Predictor CNNs and Shape Synthesis (CVPR 2017 · similarity 0.87)
- SurfNet: Generating 3D Shape Surfaces Using Deep Residual Networks (CVPR 2017 · similarity 0.87)
- Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling (NeurIPS 2016 · similarity 0.86)
- Deep Optimized Priors for 3D Shape Modeling and Reconstruction (CVPR 2021 · similarity 0.85)
- $\texttt{NAISR}$: A 3D Neural Additive Model for Interpretable Shape Representation (ICLR 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.