Research map / Label-efficient and robust learning / Domain adaptation and generalization
Multi-Domain Learning for Updating Face Anti-Spoofing Models
Xiao Guo, Yaojie Liu, Anil Jain, Xiaoming Liu
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 LSpoof (spoof region estimator loss) (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 76.3 -> 82.1 | Improved |
| add LSpoof (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | ACER source: 0.9% ACER | Improved |
| add LSpoof (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | TPR@FPR=0.5% target: 91.1 -> 92.0 | No clear effect |
| add adversarial losses LS and LT (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 80.8 -> 82.1 | Improved |
| add adversarial losses LS and LT (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | TPR@FPR=0.5% target: 91.3 -> 92.0 | Improved |
| add regularization term LT (ablation) | FAS-wrapper (PhySTD) · FASMD (average of 4 protocols) | ACER target: 0.4% ACER | Improved |
| add LSpoof to FAS-wrapper (ablation) | PhySTD · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 76.3 -> 82.1 | Improved |
| add LSpoof to FAS-wrapper (ablation) | Auxi.-CNN · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 69.0 -> 78.0 | Improved |
| add adversarial losses LT + LS (ablation) | CDCN · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 78.6 -> 80.0 | Improved |
| add adversarial losses LT + LS (ablation) | Auxi.-CNN · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 75.0 -> 78.0 | Improved |
| use FAS-wrapper instead of naive fine-tuning (ablation) | CDCN · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 72.2 -> 80.0 | Improved |
| use FAS-wrapper instead of naive fine-tuning (ablation) | PhySTD · FASMD (average of 4 protocols) | TPR@FPR=0.5% source: 74.8 -> 82.1 | Improved |
Closest papers by meaning
- Domain-Hallucinated Updating for Multi-Domain Face Anti-spoofing (AAAI 2024 · similarity 0.93)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.91)
- Gradient Alignment for Cross-Domain Face Anti-Spoofing (CVPR 2024 · similarity 0.91)
- One-Class Face Anti-spoofing via Spoof Cue Map-Guided Feature Learning (CVPR 2024 · similarity 0.90)
- mmFAS: Multimodal Face Anti-Spoofing Using Multi-Level Alignment and Switch-Attention Fusion (AAAI 2025 · similarity 0.89)
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.89)
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.
