AI-generated image detection on AIGIBench WFIR
94.8AccuracyC2P-CLIP
Evaluation results
| Method | Links | ||
|---|---|---|---|
| C2P-CLIPevaluation_strategy=cross-model2025.01 | 94.8 | 99.5 | |
| UnivFDevaluation_strategy=cross-model2025.01 | 91 | 97.8 | |
| AIDEevaluation_strategy=cross-model2025.01 | 90.3 | 97 | |
| FatFormerevaluation_strategy=cross-model2025.01 | 88.2 | 98.5 | |
| LEGOVenue=-2026.05 | 86.1 | 95.8 | |
| LaDeDaevaluation_strategy=cross-model2025.01 | 84.5 | 91.3 | |
| NPRevaluation_strategy=cross-model2025.01 | 77.8 | 77 | |
| CNN-Spotevaluation_strategy=cross-model2025.01 | 62.1 | 89.4 | |
| FreqNetevaluation_strategy=cross-model2025.01 | 58.5 | 48.9 | |
| LGradevaluation_strategy=cross-model2025.01 | 55.6 | 53.1 | |
| DFFreqevaluation_strategy=cross-model2025.01 | 52.2 | 46.8 |