AI-generated image detection on SDXL-Turbo (test)
74.44AccuracyFreDect
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FreDectTraining Dataset=ProGAN2026.01 | 74.44 | 81.73 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 53.59 | 76.42 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 53.41 | 75.13 | |
| FatFormerTraining Dataset=ProGAN2026.01 | 51.69 | 50.63 | |
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 50.64 | 76.67 | |
| SAFETraining Dataset=ProGAN2026.01 | 49.9 | 51.07 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 49.75 | 64.03 | |
| NPRTraining Dataset=ProGAN2026.01 | 49.5 | 36.82 | |
| LNPTraining Dataset=ProGAN2026.01 | 40.6 | 31.86 | |
| GramNetTraining Dataset=ProGAN2026.01 | 32.32 | 33.18 | |
| LGradTraining Dataset=ProGAN2026.01 | 30.96 | 31.36 |