Research map / Label-efficient and robust learning / Recognition and long-tail
BioMetricNet: deep unconstrained face verification through learning of metrics regularized onto Gaussian distributions
Arslan Ali, Matteo Testa, Tiziano Bianchi, Enrico Magli
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 |
|---|---|---|---|
| increase feature dimension d from 128 to 512 (ablation) | BioMetricNet · LFW | verification accuracy (%): 99.47 -> 99.80 | Improved |
| increase feature dimension d from 128 to 512 (ablation) | BioMetricNet · CPLFW | verification accuracy (%): 94.89 -> 95.60 | Improved |
| increase feature dimension d from 512 to 1024 (ablation) | BioMetricNet · LFW | verification accuracy (%): 99.80 -> 99.63 | Hurt |
| increase feature dimension d from 512 to 1024 (ablation) | BioMetricNet · CPLFW | verification accuracy (%): 95.60 -> 95.25 | Hurt |
| increase latent dimension p from 1 to 16 (ablation) | BioMetricNet · LFW | verification accuracy (%): 99.80 -> 99.72 | Hurt |
| increase latent dimension p from 1 to 16 (ablation) | BioMetricNet · CPLFW | verification accuracy (%): 95.60 -> 95.43 | Hurt |
| increase latent dimension p from 1 to 16 (ablation) | BioMetricNet · CFP-FP | verification accuracy (%): 99.35 -> 99.47 | Improved |
| increase latent dimension p from 1 to 8 (ablation) | BioMetricNet · CPLFW | verification accuracy (%): 95.60 -> 95.13 | Hurt |
| increase latent dimension p from 1 to 3 (ablation) | BioMetricNet · CALFW | verification accuracy (%): 97.07 -> 97.02 | No clear effect |
| replace fixed analytic metric with learned metric regularized to Gaussians, versus ArcFace | BioMetricNet vs ArcFace · CPLFW | verification accuracy (%): 92.08 -> 95.60 | Improved |
| replace fixed analytic metric with learned metric regularized to Gaussians, versus ArcFace | BioMetricNet vs ArcFace · LFW | verification accuracy (%): 99.82 vs 99.80 | No clear effect |
| replace fixed analytic metric with learned metric regularized to Gaussians, versus ArcFace | BioMetricNet vs ArcFace · CALFW | verification accuracy (%): 95.45 -> 97.07 | Improved |
Later papers that compared with this method
1 later accepted paper reports BioMetricNet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Killing Two Birds With One Stone: Efficient and Robust Training of Face Recognition CNNs by Partial FC (CVPR 2022 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ArcFace — ArcFace: Additive Angular Margin Loss for Deep Face Recognition (CVPR 2019) · better on 5 of 6 shared results
- CosFace — CosFace: Large Margin Cosine Loss for Deep Face Recognition (CVPR 2018) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- An Efficient Joint Formulation for Bayesian Face Verification (TPAMI 2016 · similarity 0.87)
- Large Scale Similarity Learning Using Similar Pairs for Person Verification (AAAI 2016 · similarity 0.87)
- Temporal Non-Volume Preserving Approach to Facial Age-Progression and Age-Invariant Face Recognition (ICCV 2017 · similarity 0.86)
- Towards Age-Invariant Face Recognition (TPAMI 2020 · similarity 0.86)
- Probabilistic Elastic Part Model: A Pose-Invariant Representation for Real-World Face Verification (TPAMI 2017 · similarity 0.86)
- 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.
