AI-generated image detection on FakeForm Depth Map
92.8AccuracyCoDA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CoDABackbone=ResNet2026.05 | 92.8 | 97.7 | |
| FatFormerBackbone=ViT2026.05 | 91.2 | 96.8 | |
| AIGI-HolmesBackbone=ViT2026.05 | 85.4 | 91.8 | |
| NPRBackbone=ResNet2026.05 | 80.1 | 94.1 | |
| FreqNetBackbone=ResNet2026.05 | 76.3 | 93.2 | |
| AIDEBackbone=ResNeXt2026.05 | 68.5 | 92.4 | |
| Humans2026.05 | 65.3 | 75.3 | |
| UFD2026.05 | 43.3 | 47.5 | |
| SAFEBackbone=ResNet2026.05 | 21.5 | 59.7 | |
| CNNDetBackbone=Xception2026.05 | 20.3 | 45.3 |