AI-generated image detection on GLIDE generator (test)
92.1AccuracyFatFormer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FatFormerTraining Dataset=ProGAN2026.01 | 92.1 | 97.88 | |
| SAFETraining Dataset=ProGAN2026.01 | 88.5 | 95.1 | |
| NPRTraining Dataset=ProGAN2026.01 | 81.7 | 91.46 | |
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 70.95 | 85.13 | |
| LGradTraining Dataset=ProGAN2026.01 | 67 | 75.95 | |
| LNPTraining Dataset=ProGAN2026.01 | 64.4 | 70.02 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 58.45 | 74.76 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 57.85 | 83.78 | |
| GramNetTraining Dataset=ProGAN2026.01 | 55.15 | 53.46 | |
| FreDectTraining Dataset=ProGAN2026.01 | 52.8 | 51.77 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 51.15 | 59.88 |