Frame-level Face Forgery Detection on Celeb-DF v2
0.9Frame-level AUCForensics Adapter
Evaluation results
| Method | Links | |
|---|---|---|
| Forensics AdapterTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.9 | |
| SFAMTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.885 | |
| EffortTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.833 | |
| UCFTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.772 | |
| CLIPTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.754 | |
| Efficientnet-B4Training Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.75 | |
| RECCETraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.741 | |
| XceptionTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.74 | |
| F3NetTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.729 | |
| FFDTraining Dataset=FaceForensics++, Evaluation Level=frame-level2026.04 | 0.687 |