Research map / Video / Video segmentation and tracking
FakeCatcher: Detection of Synthetic Portrait Videos using Biological Signals
Umur Aybars Çiftçi, İlke Demir, Lijun Yin
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 biological-signal PPG maps instead of raw frames (ablation) | CNN · Deep Fakes Dataset | accuracy: 46.89 -> 87.62 | Improved |
| add spectral power density encoding to PPG maps (ablation) | CNN · Face Forensics | video accuracy: 90.66 -> 96 | Improved |
| add spectral power density encoding to PPG maps (ablation) | CNN · Deep Fakes Dataset | video accuracy: 82.69 -> 91.07 | Improved |
| replace SVM with CNN on PPG maps (ablation) | FakeCatcher · Face Forensics | video accuracy: 82.55 -> 96 | Improved |
| add weighted probabilistic voting instead of majority voting (ablation) | SVM classifier · Face Forensics | video accuracy: 78.18 -> 82.55 | Improved |
| add differences between consecutive frames to thresholding (ablation) | pairwise statistic · Face Forensics pairs | accuracy: 65 -> 75.69 | Improved |
| add cross power spectral density of coherence signals (ablation) | pairwise statistic · Face Forensics pairs | accuracy: 97.28 -> 99.39 | Improved |
| add signal normalization (ablation) | pairwise statistic · Face Forensics pairs | accuracy: 99.39 -> 87.34 | Hurt |
| use default ROI instead of whole face (ablation) | SVM classifier · Face Forensics w=128 | video accuracy: 74.23 -> 82.55 | Improved |
| increase segment length from 128 to 600 frames (ablation) | SVM classifier · Face Forensics | video accuracy: 82.55 -> 68.42 | Hurt |
| increase Gaussian blur kernel from 3x3 to 11x11 (ablation) | FakeCatcher · Celeb-DF | accuracy: 91.31 -> 65.44 | Hurt |
| restrict to heart-rate frequency band 1.05-3.04 Hz (ablation) | pairwise statistic · Face Forensics | accuracy: 95.15 -> 88.51 | Hurt |
Closest papers by meaning
- Spot the Fake: Large Multimodal Model-Based Synthetic Image Detection with Artifact Explanation (NeurIPS 2025 · similarity 0.89)
- Community Forensics: Using Thousands of Generators to Train Fake Image Detectors (CVPR 2025 · similarity 0.88)
- FakeDiffer: Distributional Disparity Learning on Differentiated Reconstruction for Face Forgery Detection (AAAI 2025 · similarity 0.87)
- CO-SPY: Combining Semantic and Pixel Features to Detect Synthetic Images by AI (CVPR 2025 · similarity 0.86)
- Lips Are Lying: Spotting the Temporal Inconsistency between Audio and Visual in Lip-Syncing DeepFakes (NeurIPS 2024 · similarity 0.86)
- Identity-Aware Vision-Language Model for Explainable Face Forgery Detection (AAAI 2026 · 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.
