Research map / 3D vision / Face and shape
Enhancing Face Recognition With Self-Supervised 3D Reconstruction
Check 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 a self-supervised 3D reconstruction bypass to face recognition training | CurricularFace with ResNet-100 · IJB-B | TAR@FAR=1e-6: 42.26 -> 45.77 | Helped |
| add a self-supervised 3D reconstruction bypass to face recognition training | CurricularFace with ResNet-100 · IJB-C occlusion subset | TAR@FAR=1e-6: 70.23 -> 75.09 | Helped |
| add a self-supervised 3D reconstruction bypass to face recognition training | CurricularFace with ResNet-100 · IJB-C | TAR@FAR=1e-6: 87.46 -> 88.45 | Helped |
| add a self-supervised 3D reconstruction bypass to face recognition training | CurricularFace with ResNet-100 · IJB-C | TAR@FAR=1e-5: 93.85 -> 94.30 | Helped |
| add a self-supervised 3D reconstruction bypass to face recognition training | CurricularFace with ResNet-100 · IJB-C | TAR@FAR=1e-4: 96.20 -> 96.20 | No clear effect |
| add a self-supervised 3D reconstruction bypass to face recognition training | face representation for attribute recognition · LFWA | recognition accuracy: 78.68 -> 84.08 | Helped |
| add self-supervised 3D reconstruction pretraining (ablation) | face representations for LFWA attribute recognition · LFWA, 10% training samples | recognition accuracy: 80.24 -> 83.15 | Helped |
| add supervised face recognition pretraining (ablation) | face representations for LFWA attribute recognition · LFWA, 10% training samples | recognition accuracy: 80.24 -> 84.62 | Helped |
| combine self-supervised 3D reconstruction and supervised face recognition pretraining (ablation) | face representations for LFWA attribute recognition · LFWA, 10% training samples | recognition accuracy: 84.62 -> 85.33 | Helped |
| combine self-supervised 3D reconstruction and supervised face recognition pretraining (ablation) | face representations for LFWA attribute recognition · LFWA, 100% training samples | recognition accuracy: 86.43 -> 86.39 | Hurt |
Closest papers by meaning
- Self-Supervised Multi-Level Face Model Learning for Monocular Reconstruction at Over 250 Hz (CVPR 2018 · similarity 0.89)
- FML: Face Model Learning From Videos (CVPR 2019 · similarity 0.89)
- Learning From Millions of 3D Scans for Large-Scale 3D Face Recognition (CVPR 2018 · similarity 0.88)
- InverseFaceNet: Deep Monocular Inverse Face Rendering (CVPR 2018 · similarity 0.88)
- FaceMe: Robust Blind Face Restoration with Personal Identification (AAAI 2025 · similarity 0.85)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.85)
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.