Research map / 3D vision / Face and shape
Geometric Disentanglement for Generative Latent Shape Models
Tristan Aumentado-Armstrong, Stavros Tsogkas, Allan Jepson, Sven Dickinson
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 Jacobian penalty (ablation) | GDVAE · SMAL retrieval | E_beta via z_I (lower better): 0.734 -> 0.645 | Improved |
| add covariance penalty (ablation) | GDVAE · SMAL retrieval | E_beta via z_I (lower better): 0.656 -> 0.645 | Improved |
| add Jacobian and covariance penalties (ablation) | GDVAE · SMAL retrieval | E_beta via z_I (lower better): 0.891 -> 0.645 | Improved |
| add Jacobian penalty (ablation) | GDVAE · SMPL retrieval | E_theta via z_E (lower better): 0.812 -> 0.709 | Improved |
| add covariance penalty (ablation) | GDVAE · SMPL retrieval | E_theta via z_E (lower better): 0.836 -> 0.709 | Improved |
| add covariance penalty (ablation) | GDVAE · SMPL retrieval | E_beta via z_I (lower better): 0.905 -> 0.928 | Hurt |
| add Jacobian penalty (ablation) | GDVAE · SMPL retrieval | E_beta via z_I (lower better): 0.895 -> 0.928 | Hurt |
| add covariance penalty (ablation) | GDVAE · SMAL retrieval | E_theta via z_E (lower better): 0.969 -> 0.983 | Hurt |
| add Jacobian and covariance penalties (ablation) | GDVAE · SMAL retrieval | E_beta via z (lower better): 0.661 -> 0.743 | Hurt |
| add Jacobian and covariance penalties (ablation) | GDVAE · SMPL retrieval | E_beta via z (lower better): 0.888 -> 0.922 | Hurt |
| increase total correlation penalty (ablation) | GDVAE · MNIST | reconstruction log-likelihood | Hurt |
| add rotational consistency penalty on X representations (ablation) | GDVAE · MNIST/Dyna | reconstruction error | Hurt |
Closest papers by meaning
- Disentangling Geometric Deformation Spaces in Generative Latent Shape Models (IJCV 2023 · similarity 0.94)
- 3D Shape Variational Autoencoder Latent Disentanglement via Mini-Batch Feature Swapping for Bodies and Faces (CVPR 2022 · similarity 0.91)
- Shape-Pose Disentanglement Using SE(3)-Equivariant Vector Neurons (ECCV 2022 · similarity 0.89)
- Learning Representations and Generative Models for 3D Point Clouds (ICML 2018 · similarity 0.89)
- Disentangled3D: Learning a 3D Generative Model With Disentangled Geometry and Appearance From Monocular Images (CVPR 2022 · similarity 0.89)
- Discrete Point Flow Networks for Efficient Point Cloud Generation (ECCV 2020 · similarity 0.89)
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.
