AI-generated image detection on StyleGAN2 generator (test)
99.22AccuracyFatFormer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FatFormerTraining Dataset=ProGAN2026.01 | 99.22 | 99.89 | |
| SAFETraining Dataset=ProGAN2026.01 | 98.92 | 99.94 | |
| NPRTraining Dataset=ProGAN2026.01 | 98.41 | 99.97 | |
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 93.6 | 99.72 | |
| LNPTraining Dataset=ProGAN2026.01 | 93.09 | 99.2 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 86.25 | 99.06 | |
| LGradTraining Dataset=ProGAN2026.01 | 86.11 | 98.26 | |
| GramNetTraining Dataset=ProGAN2026.01 | 85.86 | 99.37 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 80.88 | 98.75 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 70.76 | 97.71 | |
| FreDectTraining Dataset=ProGAN2026.01 | 66.17 | 82.53 |