Manipulation detection on FakeAVCeleb FVFA-WL
100AP (%)AD DFD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AD DFDModality=Audio-Visual, Pretrained dataset=Kinetics, Supervision Protocol=Supervised2023.01 | 100 | 100 | |
| AV-DFDModality=AV2024.06 | 100 | 100 | |
| AVFFModality=AV2024.06 | 99.4 | 99.8 | |
| LipForensicsModality=Visual, Pretrained dataset=LRW, Supervision Protocol=Supervised2023.01 | 98.6 | 98.7 | |
| LipForensicsModality=V2024.06 | 98.6 | 98.7 | |
| FTCNModality=Visual, Pretrained dataset=None, Supervision Protocol=Supervised2023.01 | 95.6 | 96.5 | |
| FTCNModality=V2024.06 | 95.6 | 96.5 | |
| Audio-Visual Anomaly Detection (Ours)Modality=Audio-Visual, Pretrained dataset=LRS2, Supervision Protocol=Unsupervised2023.01 | 93.8 | 94.1 | |
| AVADModality=AV, Training Set=LRS22024.06 | 93.8 | 94.1 | |
| RealForensicsModality=Visual, Pretrained dataset=LRW, Supervision Protocol=Supervised2023.01 | 93.4 | 96.7 | |
| RealForensicsModality=V2024.06 | 93.4 | 96.7 | |
| Audio-Visual Anomaly Detection (Ours)Modality=Audio-Visual, Pretrained dataset=LRS3, Supervision Protocol=Unsupervised2023.01 | 91.4 | 93.1 | |
| AVADModality=AV, Training Set=LRS32024.06 | 91.4 | 93.1 | |
| XceptionModality=Visual, Pretrained dataset=ImageNet, Supervision Protocol=Supervised2023.01 | 91 | 91 | |
| XceptionModality=V2024.06 | 91 | 91 | |
| AVBYOLModality=Audio-Visual, Pretrained dataset=LRW, Supervision Protocol=Unsupervised2023.01 | 73.2 | 61 | |
| VQ-GANModality=Visual, Pretrained dataset=LRS2, Supervision Protocol=Unsupervised2023.01 | 62.4 | 56.9 |