AI-generated image detection on AIGIBench BLIP
99.2AccuracyDFFreq
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DFFreqevaluation_strategy=cross-model2025.01 | 99.2 | 99.7 | |
| NPRevaluation_strategy=cross-model2025.01 | 94 | 97 | |
| FreqNetevaluation_strategy=cross-model2025.01 | 93.8 | 99.9 | |
| LaDeDaevaluation_strategy=cross-model2025.01 | 93.4 | 96.7 | |
| AIDEevaluation_strategy=cross-model2025.01 | 92.7 | 96.7 | |
| C2P-CLIPevaluation_strategy=cross-model2025.01 | 84.9 | 93.8 | |
| FatFormerevaluation_strategy=cross-model2025.01 | 82.8 | 92.3 | |
| LEGOVenue=-2026.05 | 73.3 | 89.7 | |
| UnivFDevaluation_strategy=cross-model2025.01 | 57.8 | 68.9 | |
| LGradevaluation_strategy=cross-model2025.01 | 56.4 | 57.8 | |
| CNN-Spotevaluation_strategy=cross-model2025.01 | 49.8 | 44.7 |