Research map / Image generation / Image restoration
Face De-Spoofing: Anti-Spoofing via Noise Modeling
Amin Jourabloo, Yaojie Liu, 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 magnitude loss (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 5.24 -> 1.50 | Improved |
| add 0/1 map loss (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 2.34 -> 1.50 | Improved |
| add repetitive loss (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 1.50 | No clear effect |
| add repetitive loss on 1080P cheek-region input (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 (1080P cheek region) | ACER (%): cannot converge -> 2.92 | Improved |
| fuse spoof noise and depth map scores by average instead of using spoof noise alone (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.70 -> 1.50 | Improved |
| fuse spoof noise and depth map by average instead of depth map alone (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.67 -> 1.50 | Improved |
| use maximum instead of average fusion of spoof noise and depth (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 1.72 | Hurt |
| add 0/1 map score to fusion of spoof noise and depth (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 1.50 | No clear effect |
| reduce input resolution from 256x256 to 128x128 (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 3.07 | Hurt |
| reduce input resolution from 256x256 to 64x64 (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 5.27 | Hurt |
| blur input with 9x9 Gaussian filter (ablation) | De-Spoof Net · Oulu-NPU Protocol 1 | ACER (%): 1.50 -> 4.89 | Hurt |
| train on higher-resolution CASIA-MFSD and test on lower-resolution Replay-Attack versus reverse | De-Spoof Net · Cross test CASIA-MFSD / Replay-Attack | HTER (%): 28.5 -> 41.1 | Mixed |
Closest papers by meaning
- Learning Deep Models for Face Anti-Spoofing: Binary or Auxiliary Supervision (CVPR 2018 · similarity 0.91)
- On Disentangling Spoof Trace for Generic Face Anti-Spoofing (ECCV 2020 · similarity 0.90)
- Face Anti-Spoofing: Model Matters, so Does Data (CVPR 2019 · similarity 0.90)
- FaceShield: Explainable Face Anti-Spoofing with Multimodal Large Language Models (AAAI 2026 · similarity 0.88)
- One-Class Face Anti-spoofing via Spoof Cue Map-Guided Feature Learning (CVPR 2024 · similarity 0.88)
- SLIP: Spoof-Aware One-Class Face Anti-Spoofing with Language Image Pretraining (AAAI 2025 · similarity 0.87)
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.
