Video-level Deepfake Detection on AV-Deepfake1M++
96.66AUC (Video)multi-branch framework (Ours)
Evaluation results
| Method | Links | |
|---|---|---|
| multi-branch framework (Ours)2026.05 | 96.66 | |
| UMMAFormerInput Modalities=Visual+Audio2026.05 | 87.04 | |
| MM-DetInput Modalities=Text+Visual2026.05 | 82.61 | |
| BA-TFD+Input Modalities=Visual+Audio2026.05 | 55.43 | |
| BA-TFDInput Modalities=Visual+Audio2026.05 | 50.28 |