Research map / Label-efficient and robust learning / Recognition and long-tail
OneFace: One Threshold for All
Jiaheng Liu, Zhipeng Yu, Haoyu Qin, Yichao Wu, Ding Liang, Gangming Zhao, Ke Xu
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 Threshold Consistency Penalty (TCP) to ArcFace | ArcFace · RFW cross-race OTA (FAR 1e-4) | Avg TAR (%): 95.51 -> 96.08 | Improved |
| add TCP to ArcFace | ArcFace · RFW cross-race OTA (FAR 1e-4) | TAR Std: 1.107 -> 0.715 | Improved |
| add TCP to ArcFace | ArcFace · RFW cross-race OTA (FAR 1e-4) | -log10FAR Std: 0.378 -> 0.331 | Improved |
| add TCP to ArcFace | ArcFace · RFW cross-race OTA (FAR 1e-4) | Avg -log10FAR: 3.598 -> 3.571 | No clear effect |
| add TCP to ArcFace | ArcFace · RFW cross-race OTA (FAR 1e-4) | Threshold imbalance gamma: 0.038 -> 0.030 | Improved |
| add TCP to ArcFace | ArcFace · IJB-B gender OTA (FAR 1e-4) | Avg TAR (%): 94.41 -> 95.03 | Improved |
| add TCP to ArcFace | ArcFace · IJB-B gender OTA (FAR 1e-4) | TAR Std: 2.555 -> 2.335 | Improved |
| add TCP to ArcFace | ArcFace · RFW classical 1:1 (per-race thresholds), African | TAR@FAR=1e-4 (%): 92.29 -> 93.43 | Improved |
| add TCP to ArcFace (R100) | ArcFace · RFW verification accuracy (per-race thresholds) | Avg accuracy (%): 96.24 -> 97.01 | Improved |
| add TCP to ArcFace training | ArcFace · Training cost | training time ratio: 1.183 times time, 1.005 times GPU memory | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- ArcFace — ArcFace: Additive Angular Margin Loss for Deep Face Recognition (CVPR 2019) · better on 8 of 9 shared results
- CurricularFace — CurricularFace: Adaptive Curriculum Learning Loss for Deep Face Recognition (CVPR 2020) · better on 6 of 7 shared results
- GroupFace — GroupFace: Learning Latent Groups and Constructing Group-Based Representations for Face Recognition (CVPR 2020) · better on 6 of 7 shared results
- CosFace — CosFace: Large Margin Cosine Loss for Deep Face Recognition (CVPR 2018) · better on 6 of 7 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
- UniTSFace: Unified Threshold Integrated Sample-to-Sample Loss for Face Recognition (NeurIPS 2023 · similarity 0.86)
- AnchorFace: Boosting TAR@FAR for Practical Face Recognition (AAAI 2022 · similarity 0.86)
- Instance-Consistent Fair Face Recognition (TPAMI 2025 · similarity 0.85)
- Goldilocks Test Sets for Face Verification (CVPR 2026 · similarity 0.85)
- TransFace++: Rethinking the Face Recognition Paradigm With a Focus on Accuracy, Efficiency, and Security (TPAMI 2025 · similarity 0.85)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.82)
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.
