Research map / Image generation / Editing and style transfer
Generalized Face Liveness Detection via De-Fake Face Generator
Xingming Long, Jie Zhang, Shiguang Shan
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 anomalous cue via cross-attention (OA-Net) to ViT-base (ablation) | ViT-base FAS · Leave-one-out FAS (mean) | HTER: 9.68 -> 6.82 | Improved |
| use only the residual cue instead of the image (ablation) | OA-Net · Leave-one-out FAS (mean) | HTER: 9.68 -> 20.63 | Hurt |
| use dual-branch concatenation instead of cross-attention (ablation) | ViT-base · Leave-one-out FAS (mean) | HTER: 9.68 -> 9.56 | No clear effect |
| use DFG cue instead of MemAE cue (ablation) | OA-Net · Leave-one-out FAS (mean) | HTER: 7.67 -> 6.82 | Improved |
| use DFG cue instead of Deep SVDD cue (ablation) | OA-Net · Leave-one-out FAS (mean) | HTER: 8.37 -> 6.82 | Improved |
| use DFG cue instead of AnoDDPM cue (ablation) | OA-Net · Leave-one-out FAS (mean) | HTER: 8.34 -> 6.82 | Improved |
| mask the background of the anomalous cue (ablation) | DFG cue · Leave-one-out FAS (mean) | HTER: 6.82 -> 6.83 | No clear effect |
| use diffusion step 800 instead of 1000 (ablation) | DFG · Leave-one-out FAS (mean) | HTER: 8.12 -> 6.82 | Improved |
| use diffusion step 800 instead of 200 (ablation) | DFG · Leave-one-out FAS (mean) | HTER: 7.67 -> 6.82 | Improved |
| train DFG on 100% instead of 25% of real faces (ablation) | DFG · Leave-one-out FAS (mean) | HTER: 8.27 -> 6.82 | Improved |
| add 130K additional real faces to training of SA-FAS (ablation) | SA-FAS · Leave-one-out FAS (mean) | HTER: 7.83 -> 24.45 | Hurt |
| add 130K additional real faces to training of SSAN-R (ablation) | SSAN-R · Leave-one-out FAS (mean) | HTER: 9.82 -> 16.32 | Hurt |
Closest papers by meaning
- Towards Unsupervised Domain Generalization for Face Anti-Spoofing (ICCV 2023 · similarity 0.89)
- One-Class Face Anti-spoofing via Spoof Cue Map-Guided Feature Learning (CVPR 2024 · similarity 0.89)
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.89)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.89)
- From Intuition to Investigation: A Tool-Augmented Reasoning MLLM Framework for Generalizable Face Anti-Spoofing (CVPR 2026 · similarity 0.86)
- Gradient Alignment for Cross-Domain Face Anti-Spoofing (CVPR 2024 · similarity 0.86)
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.
