Research map / Label-efficient and robust learning / Recognition and long-tail
Probabilistic Face Embeddings
Yichun Shi, Anil K. Jain
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 |
|---|---|---|---|
| convert deterministic embedding to PFE with mutual likelihood matching | 64-layer CNN + AM-Softmax (CASIA-WebFace) · IJB-A TAR@FAR=0.1% | verification rate: 84.69 -> 87.58 | Improved |
| convert deterministic embedding to PFE | 64-layer CNN + A-Softmax (CASIA-WebFace) · IJB-A TAR@FAR=0.1% | verification rate: 78.54 -> 82.58 | Improved |
| convert deterministic embedding to PFE | 64-layer CNN + Softmax+Center Loss (CASIA-WebFace) · IJB-A TAR@FAR=0.1% | verification rate: 78.16 -> 80.83 | Improved |
| convert deterministic embedding to PFE | 64-layer CNN + Triplet (CASIA-WebFace) · IJB-A TAR@FAR=0.1% | verification rate: 60.82 -> 61.00 | No clear effect |
| convert deterministic embedding to PFE | CASIA-Net + Triplet (CASIA-WebFace) · IJB-A TAR@FAR=0.1% | verification rate: 54.47 -> 51.35 | Hurt |
| add uncertainty-based fusion instead of average pooling (ablation) | Baseline AM-Softmax embedding (MS-Celeb-1M) · IJB-C TAR@FAR=0.001% | TAR: 70.10 -> 83.14 | Improved |
| add mutual likelihood score matching on top of PFE fusion (ablation) | PFE fuse (MS-Celeb-1M) · IJB-C TAR@FAR=0.001% | TAR: 83.14 -> 89.64 | Improved |
| add uncertainty-based fusion instead of average pooling (ablation) | Baseline AM-Softmax embedding (MS-Celeb-1M) · IJB-A TAR@FAR=0.1% | TAR: 93.30 -> 94.59 | Improved |
| add mutual likelihood score matching on top of PFE fusion (ablation) | PFE fuse (MS-Celeb-1M) · IJB-A TAR@FAR=1.0% | TAR: 95.92 -> 97.50 | Improved |
| convert deterministic embedding to PFE with fusion and matching | 64-layer CNN + AM-Softmax (MS-Celeb-1M) · MegaFace MF1 Rank-1 | Rank-1: 79.43 -> 78.95 | No clear effect |
| add mutual likelihood score matching on top of PFE fusion (ablation) | PFE fuse (MS-Celeb-1M) · IJB-S surveillance-to-single Rank-1 | Rank-1: 53.44 -> 50.16 | Mixed |
| add mutual likelihood score matching on top of PFE fusion (ablation) | PFE fuse (MS-Celeb-1M) · IJB-S surveillance-to-single TPIR@FPIR=1% | TPIR: 10.53 -> 31.88 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- ArcFace — ArcFace: Additive Angular Margin Loss for Deep Face Recognition (CVPR 2019) · better on 0 of 1 shared result
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
- Spherical Confidence Learning for Face Recognition (CVPR 2021 · similarity 0.90)
- Data Uncertainty Learning in Face Recognition (CVPR 2020 · similarity 0.90)
- An Efficient Joint Formulation for Bayesian Face Verification (TPAMI 2016 · similarity 0.88)
- Attributes Shape the Embedding Space of Face Recognition Models (ICML 2025 · similarity 0.84)
- HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere (ICLR 2025 · similarity 0.84)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.83)
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.
