AI-generated image detection on Stable Diffusion generator v2.1 (test)
60.98AccuracyOurs (ViT-L/14)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 60.98 | 86.36 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 60.14 | 82.58 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 57.26 | 85.36 | |
| FatFormerTraining Dataset=ProGAN2026.01 | 51.8 | 51.34 | |
| SAFETraining Dataset=ProGAN2026.01 | 49.94 | 51.22 | |
| NPRTraining Dataset=ProGAN2026.01 | 49.45 | 42.2 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 48.57 | 55.76 | |
| LNPTraining Dataset=ProGAN2026.01 | 42.29 | 36.66 | |
| GramNetTraining Dataset=ProGAN2026.01 | 32.73 | 32.99 | |
| FreDectTraining Dataset=ProGAN2026.01 | 31.64 | 34.77 | |
| LGradTraining Dataset=ProGAN2026.01 | 31.45 | 33.17 |