AI-generated image detection on GauGAN generator (test)
99.95AccuracyOurs (ViT-L/14)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 99.95 | 100 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 99.47 | 99.98 | |
| FatFormerTraining Dataset=ProGAN2026.01 | 94.36 | 99.84 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 87.58 | 99.57 | |
| SAFETraining Dataset=ProGAN2026.01 | 86.09 | 94.38 | |
| FreDectTraining Dataset=ProGAN2026.01 | 80.57 | 82.86 | |
| LGradTraining Dataset=ProGAN2026.01 | 80.08 | 91.32 | |
| NPRTraining Dataset=ProGAN2026.01 | 78.7 | 79.29 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 77.68 | 87.64 | |
| LNPTraining Dataset=ProGAN2026.01 | 72.48 | 79.07 | |
| GramNetTraining Dataset=ProGAN2026.01 | 57.78 | 55.28 |