Image-level Forgery Detection on 7 datasets averaged (test)
90.9TNRTruFor
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TruForThreshold Selection=fixed2022.12 | 90.9 | 65.6 | 78.3 | — | |
| TruFor (max)Threshold Selection=best2022.12 | 90 | 57.3 | 73.6 | 0.996 | |
| TruFor (w/o c.)Threshold Selection=fixed2022.12 | 85.9 | 57.5 | 71.7 | — | |
| TruForThreshold Selection=best2022.12 | 85.1 | 72.9 | 79 | 0.38 | |
| CAT-Net v2Threshold Selection=best2022.12 | 84 | 61.8 | 72.9 | 0.955 | |
| TruFor (w/o c.)Threshold Selection=best2022.12 | 81.8 | 64 | 72.9 | 0.427 | |
| MVSS-NetThreshold Selection=best2022.12 | 80.6 | 54.4 | 67.5 | 0.957 | |
| IF-OSNThreshold Selection=best2022.12 | 76.3 | 54.8 | 65.6 | 0.968 | |
| CAT-Net v2Threshold Selection=fixed2022.12 | 41.6 | 88.2 | 64.9 | — | |
| MVSS-NetThreshold Selection=fixed2022.12 | 28.5 | 88.6 | 58.6 | — | |
| IF-OSNThreshold Selection=fixed2022.12 | 18.2 | 90.7 | 54.5 | — | |
| TruFor (max)Threshold Selection=fixed2022.12 | 10.9 | 96.7 | 53.8 | — |