Research map / Image generation / Editing and style transfer
SynFace: Face Recognition With Synthetic Data
Haibo Qiu, Baosheng Yu, Dihong Gong, Zhifeng Li, Wei Liu, Dacheng Tao
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 identity mixup during training (ablation) | SynFace Syn_10K_50 · LFW | accuracy (%): 88.98 -> 91.97 | Improved |
| add identity mixup during training (ablation) | SynFace 1K_50 · LFW | accuracy (%): 83.85 -> 87.53 | Improved |
| add identity mixup during training (ablation) | SynFace 10K_5 · LFW | accuracy (%): 88.22 -> 88.32 | No clear effect |
| increase mixup level from none to phi 0.6-1.0 (intra-class variation) (ablation) | SynFace 5K_50 · LFW | accuracy (%): 88.75 -> 90.95 | Improved |
| balance class distribution (2K_UB1 to 2K_50) (ablation) | SynFace · LFW long-tailed | accuracy (%): 83.80 -> 86.18 | Improved |
| increase identities (width) from 1K to 5K (ablation) | SynFace · LFW | accuracy (%): 83.85 -> 88.75 | Improved |
| increase identities (width) from 5K to 10K (ablation) | SynFace · LFW | accuracy (%): 88.75 -> 88.98 | No clear effect |
| increase samples per identity (depth) from 2 to 20 (ablation) | SynFace 10K · LFW | accuracy (%): 78.85 -> 89.90 | Improved |
| increase samples per identity (depth) from 20 to 50 (ablation) | SynFace 10K · LFW | accuracy (%): 89.90 -> 88.98 | No clear effect |
| vary pose and illumination rather than fixing attributes (ablation) | SynFace 5K_50 · LFW | accuracy (%): 74.55 -> 73.72 | Improved |
| add 1K real identities with 10 samples each via domain mixup | SynFace Syn_10K_50 with IM · LFW | accuracy (%): 91.97 -> 92.28 | Improved |
| add domain mixup to real data (2K ids, 10 samples) | Real_2K_10 · LFW | accuracy (%): 91.22 -> 95.78 | Improved |
Closest papers by meaning
- UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognition (ICLR 2025 · similarity 0.92)
- HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere (ICLR 2025 · similarity 0.91)
- CemiFace: Center-based Semi-hard Synthetic Face Generation for Face Recognition (NeurIPS 2024 · similarity 0.90)
- $\text{ID}^3$: Identity-Preserving-yet-Diversified Diffusion Models for Synthetic Face Recognition (NeurIPS 2024 · similarity 0.90)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.90)
- Data Synthesis with Diverse Styles for Face Recognition via 3DMM-Guided Diffusion (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.
