Research map / Label-efficient and robust learning / Recognition and long-tail
Benchmarking Algorithmic Bias in Face Recognition: An Experimental Approach Using Synthetic Faces and Human Evaluation
Hao Liang, Pietro Perona, Guha Balakrishnan
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 |
|---|---|---|---|
| compare pairs from same demographic group versus different groups (ablation) | face recognition models (ArcFace ResNet-34 on MS1MV3 and Glint360k, SphereFace SFNet-20 on VGGFace2) · synthetic EG3D face pairs | identity similarity | Hurt |
| increase pose angle away from 0 degrees (ablation) | ResNet34-MS1MV3 ArcFace · synthetic face pairs | similarity score | Hurt |
| test on Black and East Asian faces versus White faces | all three face recognition models · synthetic benchmark (FNMR vs FMR) | FNMR at fixed FMR | Hurt |
| test on Black Female faces | Model 3 (ResNet34 Glint360k) · pose changes | FNMR vs FMR | Hurt |
| vary lighting (ablation) | face recognition models · synthetic benchmark | FNMR vs FMR across demographic groups | No clear effect |
| vary age and expression instead of lighting (ablation) | face recognition models · synthetic benchmark | similarity | Mixed |
| remove hair via segmentation from images (ablation) | face recognition evaluation · synthetic benchmark | bias trends | No clear effect |
| change human identity threshold t_hcic to 0.2 or 0.4 (ablation) | FNMR vs FMR evaluation · synthetic benchmark | bias trends | No clear effect |
Closest papers by meaning
- Towards causal benchmarking of bias in face analysis algorithms (ECCV 2020 · similarity 0.93)
- The Value of AI Guidance in Human Examination of Synthetically-Generated Faces (AAAI 2023 · similarity 0.90)
- AI-Face: A Million-Scale Demographically Annotated AI-Generated Face Dataset and Fairness Benchmark (CVPR 2025 · similarity 0.88)
- HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere (ICLR 2025 · similarity 0.87)
- UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognition (ICLR 2025 · similarity 0.87)
- $\text{ID}^3$: Identity-Preserving-yet-Diversified Diffusion Models for Synthetic Face Recognition (NeurIPS 2024 · 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.
