Research map / Label-efficient and robust learning / Recognition and long-tail
Towards Universal Representation Learning for Deep Face Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 variation augmentation alone (ablation) | Baseline face recognizer · IJB-A verification, FAR=0.01% | TAR: 93.3±3.0 -> 93.9±3.5 | Helped |
| Add variation augmentation alone (ablation) | Baseline face recognizer · IJB-C verification, FAR=0.01% | TAR: 86.7 -> 90.6 | Helped |
| Add variation augmentation alone (ablation) | Baseline face recognizer · CFP-FP | Accuracy: 98.16 -> 98.35 | Helped |
| Add variation augmentation alone (ablation) | Baseline face recognizer · IJB-C verification, FAR=0.001% | TAR: 43.9 -> 47.6 | Helped |
| Add confidence-aware identification loss (ablation) | Variation-augmented model · IJB-A, FAR=0.001% | TAR: 82.42 -> 94.70 | Helped |
| Add confidence-aware identification loss (ablation) | Variation-augmented model · TinyFace | Rank1: 55.26 -> 57.11 | Helped |
| Split the embedding into multiple sub-embeddings (ablation) | Confidence-aware model · TinyFace | Rank1: 57.11 -> 57.30 | No clear effect |
| Add decorrelation loss to the multiple-sub-embedding model (ablation) | Model C · TinyFace | Rank1: 57.30 -> 61.32 | Helped |
| Add probabilistic aggregation (ablation) | Full model with decorrelated embeddings · TinyFace | Rank1: 61.32 -> 63.89 | Helped |
| Add probabilistic aggregation (ablation) | Full model with decorrelated embeddings · IJB-C verification, FAR=0.001% | TAR: 91.6 -> 95.0 | Helped |
| Add probabilistic aggregation without variation augmentation (ablation) | Model D · TinyFace | Rank1: 34.76 -> 35.05 | No clear effect |
| Add 3 augmentable variations to decorrelation (ablation) | Full framework · TinyFace | Rank1: 55.04 -> 54.99 | No clear effect |
Closest papers by meaning
- RegularFace: Deep Face Recognition via Exclusive Regularization (CVPR 2019 · similarity 0.88)
- Pose-Aware Face Recognition in the Wild (CVPR 2016 · similarity 0.87)
- Masked Face Recognition with Generative-to-Discriminative Representations (ICML 2024 · similarity 0.86)
- HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere (ICLR 2025 · similarity 0.85)
- FSFM: A Generalizable Face Security Foundation Model via Self-Supervised Facial Representation Learning (CVPR 2025 · similarity 0.84)
- LAFS: Landmark-based Facial Self-supervised Learning for Face Recognition (CVPR 2024 · similarity 0.84)
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.