Image-level manipulation detection on CASIA v1
0.839AUCMVSS-Net
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MVSS-NetDecision threshold=0.52021.04 | 0.839 | 0.615 | 0.969 | 0.752 | |
| FCNDecision threshold=0.52021.04 | 0.796 | 0.717 | 0.844 | 0.775 | |
| CR-CNNDecision threshold=0.52021.04 | 0.766 | 0.93 | 0.224 | 0.361 | |
| FRAMESelection=Learned, Fusion=Learned2026.05 | 0.741 | — | — | — | |
| TruFor2026.05 | 0.724 | — | — | — | |
| CAT-Net2026.05 | 0.678 | — | — | — | |
| XGB-Ensemble (pyIFD)Fusion=XGB-Ensemble2026.05 | 0.674 | — | — | — | |
| RF-Ensemble (pyIFD)Fusion=RF-Ensemble2026.05 | 0.658 | — | — | — | |
| ManTraNet2026.05 | 0.651 | — | — | — | |
| Heuristic-K + uniformSelection=Heuristic, Fusion=Uniform2026.05 | 0.628 | — | — | — | |
| Best single pyIFDSelection=Best single2026.05 | 0.612 | — | — | — | |
| Random-K + uniformSelection=Random, Fusion=Uniform2026.05 | 0.589 | — | — | — | |
| Uniform-all pyIFDSelection=None (all), Fusion=Uniform2026.05 | 0.574 | — | — | — | |
| MMFusion2026.05 | 0.51 | — | — | — | |
| GSR-NetDecision threshold=0.52021.04 | 0.502 | 0.994 | 0.011 | 0.022 | |
| ManrTra-NetDecision threshold=0.52021.04 | 0.141 | 1 | 0 | 0 |