AI-generated image detection on PNDM generator (test)
99.46AccuracyFatFormer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FatFormerTraining Dataset=ProGAN2026.01 | 99.46 | 99.09 | |
| LNPTraining Dataset=ProGAN2026.01 | 85.17 | 98.37 | |
| Ours (ViT-L/14)Training Dataset=ProGAN, Backbone=ViT-L/142026.01 | 84.28 | 99.31 | |
| SAFETraining Dataset=ProGAN2026.01 | 75.84 | 97.53 | |
| UnivFDTraining Dataset=ProGAN2026.01 | 73.95 | 97.31 | |
| Ours (ViT-B/32)Training Dataset=ProGAN, Backbone=ViT-B/322026.01 | 70.25 | 91.04 | |
| FreDectTraining Dataset=ProGAN2026.01 | 58.17 | 56.8 | |
| LGradTraining Dataset=ProGAN2026.01 | 55.01 | 60.74 | |
| CNNSpotTraining Dataset=ProGAN2026.01 | 51.31 | 74.61 | |
| NPRTraining Dataset=ProGAN2026.01 | 50.94 | 99.33 | |
| GramNetTraining Dataset=ProGAN2026.01 | 50.13 | 91.14 |