Research map / Video / Action recognition
Deepfake Video Detection via Facial Action Dependencies Estimation
Lingfeng Tan, Yunhong Wang, Junfu Wang, Liang Yang, Xunxun Chen, Yuanfang Guo
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 facial action unit features (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 89.30 -> 93.77 | Improved |
| Add MDGM with convolution-based update layers (ablation) | Xception with AU features on FF++ LQ · FF++ LQ | AUC: 89.30 -> 95.98 | Improved |
| Use convolution-based updates instead of the original GCN update (ablation) | MDGM on FF++ LQ · FF++ LQ | AUC: 94.97 -> 95.98 | Improved |
| Add the Action Branch (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 89.30 -> 92.72 | Improved |
| Add Pattern Mixer to the Action Branch (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 92.72 -> 94.54 | Improved |
| Add the Content Branch (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 89.30 -> 93.68 | Improved |
| Use full MDGM rather than Xception alone (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 89.30 -> 95.98 | Improved |
| Replace MDGM with two MHSA blocks (ablation) | Xception on FF++ LQ · FF++ LQ | AUC: 89.30 -> 93.32 | Improved |
| Use feature patches instead of AU features as MHSA input (ablation) | Xception with MHSA on FF++ LQ · FF++ LQ | AUC: 93.32 -> 89.99 | Hurt |
| Remove the Content Branch (ablation) | FADE trained on FF++ HQ · Celeb-DF | AUC: 74.83 -> 77.46 | Improved |
| Remove the Content Branch (ablation) | FADE trained on FF++ HQ · DFD | AUC: 98.41 -> 96.23 | Hurt |
| Integrate MDGM into Xception | Frame-level detector on FF++ LQ · FF++ LQ | AUC: +6.68 | Improved |
Closest papers by meaning
- Delving into the Local: Dynamic Inconsistency Learning for DeepFake Video Detection (AAAI 2022 · similarity 0.89)
- Generalizing Deepfake Video Detection with Plug-and-Play: Video-Level Blending and Spatiotemporal Adapter Tuning (CVPR 2025 · similarity 0.88)
- Towards a Universal Synthetic Video Detector: From Face or Background Manipulations to Fully AI-Generated Content (CVPR 2025 · similarity 0.87)
- Exploiting Style Latent Flows for Generalizing Deepfake Video Detection (CVPR 2024 · similarity 0.86)
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (CVPR 2025 · similarity 0.86)
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.85)
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.
