Real-vs-Fake detection on OSFFD (test)
87.22FSDLED
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DLEDBackbone=CLIP ViT-B/16, Precision=fp162025.12 | 87.22 | 85.93 | 83.52 | 84.97 | 85.41 | |
| NPRBackbone=Modified ResNet2025.12 | 80.76 | 75.73 | 77.67 | 77.21 | 77.84 | |
| D3Backbone=CLIP ViT-B/162025.12 | 78.56 | 77 | 79.67 | 79.81 | 78.76 | |
| CLIPingBackbone=CLIP ViT-B/162025.12 | 78.46 | 77.15 | 79.58 | 81.09 | 79.07 | |
| UnivFDBackbone=CLIP ViT-B/162025.12 | 77.64 | 76.83 | 79.33 | 81.31 | 78.78 | |
| SPSLBackbone=Xception2025.12 | 72.29 | 65.87 | 70.34 | 69.57 | 69.52 | |
| Xception2025.12 | 71.73 | 67.98 | 67.19 | 67.49 | 68.6 | |
| UCFBackbone=Xception2025.12 | 71.1 | 64.78 | 65.18 | 67.98 | 67.26 | |
| SIABackbone=EfficientNet-B42025.12 | 69.45 | 64.13 | 66.91 | 64.64 | 66.28 | |
| CLIP Zero-ShotBackbone=CLIP ViT-B/16, Inference mode=Zero-shot2025.12 | 52.96 | 53.2 | 53.12 | 56.62 | 53.97 | |
| SBIBackbone=EfficientNet-B42025.12 | 50.13 | 50.36 | 50.07 | 49.96 | 50.13 | |
| OC-FakeDectBackbone=Variational Autoencoder2025.12 | 48.09 | 48.45 | 48.18 | 47.16 | 47.97 |