Research map / 3D vision / Face and shape
Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency
Jiaxiang Shang, Tianwei Shen, Shiwei li, Lei Zhou, Mingmin Zhen, Tian Fang, Long Quan
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 multi-view geometry consistency losses to 2D feature losses (ablation) | face reconstruction network · AFLW2000-3D yaw 60-90 | NME: 4.31 -> 3.76 | Improved |
| add multi-view losses (ablation) | face network · AFLW2000-3D mean | NME: 3.45 -> 3.20 | Improved |
| add pixel consistency loss with covisible map (ablation) | baseline · MICC Florence cooperative | RMSE (mm): 1.83 -> 1.80 | Improved |
| add depth consistency loss with covisible map (ablation) | baseline · MICC Florence cooperative | RMSE (mm): 1.83 -> 1.77 | Improved |
| add epipolar loss (ablation) | baseline · MICC Florence outdoor | RMSE (mm): 1.81 -> 1.77 | Improved |
| add epipolar loss (ablation) | baseline · MICC Florence indoor | RMSE (mm): 1.82 -> 1.81 | No clear effect |
| add pixel and depth losses together (ablation) | baseline with covisible map · MICC Florence indoor | RMSE (mm): 1.82 -> 1.81 | No clear effect |
| add covisible map to pixel+depth+epipolar losses (ablation) | MGCNet · MICC Florence cooperative | RMSE (mm): 1.80 -> 1.73 | Improved |
| add covisible map (ablation) | MGCNet · MICC Florence outdoor | RMSE (mm): 1.82 -> 1.75 | Improved |
| add all multi-view losses and covisible map (ablation) | baseline · MICC Florence cooperative | RMSE (mm): 1.83 -> 1.73 | Improved |
| use MGCNet instead of Deng et al. | 3D face reconstruction · MICC Florence cooperative | RMSE (mm): 1.83 -> 1.73 | Improved |
| use MGCNet instead of PRN | face alignment · AFLW2000-3D mean | NME: 3.62 -> 3.20 | Improved |
Closest papers by meaning
- FML: Face Model Learning From Videos (CVPR 2019 · similarity 0.90)
- Semi-Supervised Monocular 3D Face Reconstruction With End-to-End Shape-Preserved Domain Transfer (ICCV 2019 · similarity 0.89)
- Deep Facial Non-Rigid Multi-View Stereo (CVPR 2020 · similarity 0.89)
- Self-Supervised Multi-Level Face Model Learning for Monocular Reconstruction at Over 250 Hz (CVPR 2018 · similarity 0.89)
- Self-Supervised 3D Human Mesh Recovery from a Single Image with Uncertainty-Aware Learning (AAAI 2024 · similarity 0.87)
- OFER: Occluded Face Expression Reconstruction (CVPR 2025 · similarity 0.87)
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.
