AI-generated image detection on FakeForm Crayon
80.3AccuracyCoDA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CoDABackbone=ResNet2026.05 | 80.3 | 92 | |
| FatFormerBackbone=ViT2026.05 | 78.5 | 90.7 | |
| AIGI-HolmesBackbone=ViT2026.05 | 75.2 | 82.5 | |
| UFD2026.05 | 73.8 | 81.7 | |
| NPRBackbone=ResNet2026.05 | 71.5 | 80.2 | |
| Humans2026.05 | 65.3 | 73.5 | |
| AIDEBackbone=ResNeXt2026.05 | 61.4 | 81.8 | |
| SAFEBackbone=ResNet2026.05 | 59.3 | 63.4 | |
| FreqNetBackbone=ResNet2026.05 | 58.3 | 61.6 | |
| CNNDetBackbone=Xception2026.05 | 56.2 | 59.8 |