Research map / 3D vision / Face and shape
Learning Neural Proto-Face Field for Disentangled 3D Face Modeling in the Wild
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 |
|---|---|---|---|
| add uncertainty-aware identity aggregation instead of average pooling (ablation) | NPF · MICC / Photoface / CelebAMaskHQ | p2p (mm): 1.87 -> 1.77 | Helped |
| add deformation modeling (ablation) | NPF · MICC / Photoface / CelebAMaskHQ | p2p (mm): 2.25 -> 1.77 | Helped |
| add the global face representation encoder Φe (ablation) | NPF deformation modeling · MICC / Photoface / CelebAMaskHQ | p2p (mm): 2.01 -> 1.77 | Helped |
| add multi-image warm-up before target-image fitting (ablation) | NPF fitting · MICC / Photoface / CelebAMaskHQ | p2p (mm): 1.84 -> 1.77 | Helped |
| add geometry consistency loss Lgc (ablation) | NPF fitting · MICC / Photoface / CelebAMaskHQ | p2p (mm): 1.80 -> 1.77 | Helped |
| add appearance consistency loss Lac (ablation) | NPF fitting · MICC / Photoface / CelebAMaskHQ | p2p (mm): 1.79 -> 1.77 | Helped |
| fit using a multi-image prior rather than single-image PTI | EG3D fitting · MICC / Photoface / CelebAMaskHQ | p2p (mm): 2.35 -> 1.77 | Helped |
| use the unfit face prototype rather than the full fitted method (ablation) | NPF · MICC / Photoface / CelebAMaskHQ | p2p (mm): 1.96 -> 1.77 | Helped |
| use NPF rather than DECA | 3D face reconstruction · MICC indoor | p2p (mm): 1.69 -> 1.75 | Hurt |
| use NPF rather than DECA | 3D face reconstruction · MICC outdoor | p2p (mm): 1.70 -> 1.81 | Hurt |
| use NPF rather than LAP | 3D face reconstruction · FG3D LQ | p2p (mm): 2.18 -> 2.15 | Helped |
| use NPF rather than EG3D + PTI | 3D face reconstruction · CelebAMask-HQ [±40°, ±60°] | IDE: 0.347 -> 0.287 | Helped |
Closest papers by meaning
- High-Fidelity Facial Avatar Reconstruction From Monocular Video With Generative Priors (CVPR 2023 · similarity 0.88)
- Learning Complete 3D Morphable Face Models From Images and Videos (CVPR 2021 · similarity 0.88)
- Regressing Robust and Discriminative 3D Morphable Models With a Very Deep Neural Network (CVPR 2017 · similarity 0.88)
- Learning To Aggregate and Personalize 3D Face From In-the-Wild Photo Collection (CVPR 2021 · similarity 0.87)
- 3D Facial Expressions through Analysis-by-Neural-Synthesis (CVPR 2024 · similarity 0.87)
- Learning Dense Correspondence for NeRF-Based Face Reenactment (AAAI 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.