AI Image Detection on AIGIBench (OOD)
98AUROCFiSeR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FiSeR2026.05 | 98 | 93.75 | |
| FiSeRCoarse-grained loss ablation=w/o2026.05 | 93.15 | 67.54 | |
| FiSeRFine-grained loss ablation=w/o2026.05 | 88.83 | 37.64 | |
| Effort2026.05 | 87.91 | 52.9 | |
| UniFD2026.05 | 86.67 | 36.57 | |
| C2P-CLIP2026.05 | 85.85 | 49.89 | |
| CLIPDetection2026.05 | 73.01 | 39.3 | |
| SPAI2026.05 | 72.1 | 12.78 | |
| DIRE2026.05 | 71.11 | 42.19 | |
| ResNet-50Backbone=ResNet-502026.05 | 69.7 | 28.81 | |
| SAFE2026.05 | 68.61 | 19.26 | |
| Gram-Net2026.05 | 65.87 | 26.82 | |
| LGrad2026.05 | 65.61 | 24.91 | |
| AIDE2026.05 | 63.48 | 29.62 | |
| LOTA2026.05 | 61.06 | 8.33 | |
| LASTED2026.05 | 57.75 | 15.55 | |
| FreqNet2026.05 | 54.58 | 7.3 | |
| CNNDetection2026.05 | 54.36 | 8.93 | |
| NPR2026.05 | 42.15 | 5.81 |