AI-generated image detection on DALL-E generator (test)
98.5AccuracySAFE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SAFETraining Dataset=ProGAN2026.01 | 98.5 | 99.59 | |
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 95.95 | 99.91 | |
| FatFormerTraining Dataset=ProGAN2026.01 | 95 | 97.85 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 92.1 | 99.11 | |
| GramNetTraining Dataset=ProGAN2026.01 | 87.8 | 98.81 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 87.3 | 94.47 | |
| NPRTraining Dataset=ProGAN2026.01 | 86.05 | 98.54 | |
| LGradTraining Dataset=ProGAN2026.01 | 85.75 | 95.6 | |
| LNPTraining Dataset=ProGAN2026.01 | 83.6 | 94.9 | |
| FreDectTraining Dataset=ProGAN2026.01 | 81.55 | 94.98 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 57.1 | 83.17 |