AI Image Detection on GenImage (OOD)
99.84AUROCFiSeR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FiSeR2026.05 | 99.84 | 99.55 | |
| LGrad2026.05 | 99.29 | 97.31 | |
| FiSeRFine-grained loss ablation=w/o2026.05 | 99.24 | 97.45 | |
| CLIPDetection2026.05 | 97.74 | 91.28 | |
| DIRE2026.05 | 97.59 | 87.72 | |
| FiSeRCoarse-grained loss ablation=w/o2026.05 | 96.71 | 82.62 | |
| AIDE2026.05 | 95.6 | 88.94 | |
| C2P-CLIP2026.05 | 94.71 | 94.69 | |
| SAFE2026.05 | 94.22 | 77.5 | |
| Effort2026.05 | 94.02 | 92.39 | |
| ResNet-50Backbone=ResNet-502026.05 | 91.49 | 60.31 | |
| Gram-Net2026.05 | 90.5 | 53.37 | |
| LASTED2026.05 | 85.26 | 23.35 | |
| SPAI2026.05 | 84.1 | 30.24 | |
| UniFD2026.05 | 80.76 | 33.99 | |
| LOTA2026.05 | 67.61 | 16.87 | |
| FreqNet2026.05 | 66.09 | 12.68 | |
| NPR2026.05 | 60.41 | 4.29 | |
| CNNDetection2026.05 | 43.96 | 0.61 |