AI Image Detection on Chameleon (OOD)
96.35AUROCFiSeR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FiSeR2026.05 | 96.35 | 86.38 | |
| FiSeRFine-grained loss ablation=w/o2026.05 | 90.08 | 70.01 | |
| FiSeRCoarse-grained loss ablation=w/o2026.05 | 88 | 38.89 | |
| DIRE2026.05 | 81.35 | 25.18 | |
| SPAI2026.05 | 78.89 | 34.41 | |
| LGrad2026.05 | 74.23 | 11.34 | |
| NPR2026.05 | 68.96 | 22.77 | |
| Gram-Net2026.05 | 67.77 | 9.78 | |
| SAFE2026.05 | 67.73 | 14.36 | |
| C2P-CLIP2026.05 | 64.18 | 32.54 | |
| ResNet-50Backbone=ResNet-502026.05 | 63.65 | 5.77 | |
| UniFD2026.05 | 62.66 | 8.57 | |
| LASTED2026.05 | 61.94 | 10.68 | |
| Effort2026.05 | 61.69 | 8.05 | |
| CLIPDetection2026.05 | 60.81 | 0.36 | |
| FreqNet2026.05 | 58.14 | 7.76 | |
| LOTA2026.05 | 57.67 | 9.89 | |
| AIDE2026.05 | 56.45 | 2.64 | |
| CNNDetection2026.05 | 56.04 | 5.61 |