Research map / 3D vision / Face and shape
Inequality-Constrained and Robust 3D Face Model Fitting
Evangelos Sariyanidi, Casey J. Zampella, Robert T. Schultz, Birkan Tunc
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 first inequality constraint (coefficient bounds) (ablation) | 3DI · AFLW2000-3D 2D landmarks L=68 | mean NME: +17.4% NME when omitted | Improved |
| add first inequality constraint (coefficient bounds) (ablation) | 3DI · AFLW2000-3D 3D landmarks L=68 | mean NME: +45% NME when omitted | Improved |
| add first inequality constraint (coefficient bounds) (ablation) | 3DI · Synthesized 2D landmarks L=68 | mean NME: +16.7% NME when omitted | Improved |
| add first inequality constraint (coefficient bounds) (ablation) | 3DI · Synthesized 3D landmarks L=68 | mean NME: +11.1% NME when omitted | Improved |
| add second inequality constraint (landmark detector error bounds) (ablation) | 3DI · AFLW2000-3D 2D landmarks L=68 | mean NME: +9.5% NME when omitted | Improved |
| add second inequality constraint (landmark detector error bounds) (ablation) | 3DI · AFLW2000-3D 3D landmarks L=68 | mean NME: +12.5% NME when omitted | Improved |
| add second inequality constraint (landmark detector error bounds) (ablation) | 3DI · Synthesized 2D landmarks L=68 | mean NME: +19.3% NME when omitted | Improved |
| add second inequality constraint (landmark detector error bounds) (ablation) | 3DI · Synthesized 3D landmarks L=68 | mean NME: +8.6% NME when omitted | Improved |
| use multiple frames instead of a single frame | 3DI · Synthesized 3D landmarks L=51 | mean NME: 0.034 -> 0.026 | Improved |
| use multiple frames instead of a single frame | 3DI · Synthesized 2D landmarks L=51 | mean NME: 0.025 -> 0.026 | No clear effect |
| use multiple frames instead of a single frame | 3DI · Synthesized dense reconstruction | mean NME: 0.033 -> 0.026 | Improved |
| use 3DI instead of PRNet on controlled data | 3DI Single-frame · Synthesized 2D landmarks L=51 | mean NME: 0.041 -> 0.025 | Improved |
Closest papers by meaning
- Inequality-Constrained 3D Morphable Face Model Fitting (TPAMI 2023 · similarity 0.96)
- Nonlinear 3D Face Morphable Model (CVPR 2018 · similarity 0.91)
- Regressing Robust and Discriminative 3D Morphable Models With a Very Deep Neural Network (CVPR 2017 · similarity 0.91)
- On Learning 3D Face Morphable Model from In-the-Wild Images (TPAMI 2019 · similarity 0.90)
- MVF-Net: Multi-View 3D Face Morphable Model Regression (CVPR 2019 · similarity 0.90)
- A Sparse and Locally Coherent Morphable Face Model for Dense Semantic Correspondence Across Heterogeneous 3D Faces (TPAMI 2021 · similarity 0.90)
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.
