Research map / Label-efficient and robust learning / Recognition and long-tail
Effective Presentation Attack Detection Driven by Face Related Task
Wentian Zhang, Haozhe Liu, Feng Liu, Raghavendra Ramachandra, Christoph Busch
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 face attribute editing (F.A.) features (ablation) | ResNet18 PAD baseline · Protocol-II cross-dataset | HTER: 26.90 -> 15.08 | Improved |
| add face recognition (F.R.) features (ablation) | ResNet18 PAD baseline · Protocol-II cross-dataset | HTER [O,M] to [C,I]: 25.65 -> 18.17 | Improved |
| add expression (F.E.) features (ablation) | ResNet18 PAD baseline · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 28.14 -> 16.62 | Improved |
| use GAT instead of Transformer cross-modal adapter for F.A. (ablation) | FRT-PAD · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 20.58 -> 13.18 | Improved |
| use GAT instead of CNN cross-modal adapter for F.A. (ablation) | FRT-PAD · Protocol-II cross-dataset | HTER [O,M] to [C,I]: 18.05 -> 16.98 | Improved |
| use Step-by-Step Graph instead of Dense Graph (ablation) | FRT-PAD · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 16.30 -> 13.18 | Improved |
| add F.A. features to CDC (ablation) | CDC · Protocol-II cross-dataset | HTER [O,M] to [C,I]: 28.94 -> 25.95 | Improved |
| add F.A. features to DeepPixBiS (ablation) | DeepPixBiS · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 22.45 -> 19.44 | Improved |
| add F.A. features to SSDG-R (ablation) | SSDG-R · Protocol-II cross-dataset | HTER [O,M] to [C,I]: 20.92 -> 15.63 | Improved |
| add F.A. features to VGG16 backbone (ablation) | VGG16 · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 26.50 -> 15.32 | Improved |
| add F.A. features to Swin-Transformer-tiny (ablation) | Swin-T · Protocol-II cross-dataset | HTER [C,I] to [O,M]: 35.76 -> 31.40 | Improved |
| use face detection (F.D.) features instead of F.R. (ablation) | FRT-PAD · Protocol-II cross-dataset | HTER [O,M] to [C,I]: 18.17 -> 21.47 | Hurt |
Closest papers by meaning
- Domain Generalization via Ensemble Stacking for Face Presentation Attack Detection (IJCV 2024 · similarity 0.90)
- Multi-Domain Incremental Learning for Face Presentation Attack Detection (AAAI 2024 · similarity 0.89)
- Multi-Adversarial Discriminative Deep Domain Generalization for Face Presentation Attack Detection (CVPR 2019 · similarity 0.87)
- Cross-Domain Face Presentation Attack Detection via Multi-Domain Disentangled Representation Learning (CVPR 2020 · similarity 0.86)
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.85)
- FaceShield: Explainable Face Anti-Spoofing with Multimodal Large Language Models (AAAI 2026 · 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.
