Image Manipulation Localization on Coverage
75.8F1 ScoreObjectFormer
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ObjectFormerFine-tuned=true2022.03 | 75.8 | 95.7 | — | — | — | — | — | |
| PSCCNetFine-tuned=true2022.03 | 72.3 | 94.1 | — | — | — | — | — | |
| Relay-SegProtocol=CAT2025.08 | 70.4 | — | — | — | — | — | — | |
| TinySAM w. ReViProtocol=Fine-tuned2026.04 | 70.01 | — | — | — | — | — | — | |
| Sparse-ViTProtocol=Fine-tuned2026.04 | 69.74 | — | — | — | — | — | — | |
| SegFormer w. ReViProtocol=Fine-tuned2026.04 | 66.44 | — | — | — | — | — | — | |
| FASA2026.04 | 65.37 | 95.62 | — | — | — | — | — | |
| Relay-ViTProtocol=CAT2025.08 | 64.7 | — | — | — | — | — | — | |
| IML-ViT2026.04 | 61.35 | 94.07 | — | — | — | — | — | |
| LDM w. ReViProtocol=Fine-tuned2026.04 | 60.84 | — | — | — | — | — | — | |
| Swin Transformer w. ReViProtocol=Fine-tuned2026.04 | 59.1 | — | — | — | — | — | — | |
| TruForProtocol=Fine-tuned2026.04 | 59.04 | — | — | — | — | — | — | |
| Mesorch2026.04 | 58.62 | 93.27 | — | — | — | — | — | |
| MesorchProtocol=CAT-Net protocol2025.09 | 58.6 | — | — | — | 54.4 | 59.3 | — | |
| MesorchProtocol=CAT2025.08 | 58.6 | — | — | — | — | — | — | |
| IML-ViTProtocol=Fine-tuned2026.04 | 58.14 | — | — | — | — | — | — | |
| MaskCLIP2026.04 | 56.91 | 93.82 | — | — | — | — | — | |
| Relay-SegProtocol=MVSS2025.08 | 56.9 | — | — | — | — | — | — | |
| RITAProtocol=CAT-Net protocol2025.09 | 56.6 | — | — | — | 61.8 | 64.3 | — | |
| MPCProtocol=Fine-tuned2026.04 | 56.52 | — | — | — | — | — | — | |
| SPANFine-tuned=true2022.03 | 55.8 | 93.7 | — | — | — | — | — | |
| Relay-ViTProtocol=MVSS2025.08 | 55.1 | — | — | — | — | — | — | |
| APSC-NetProtocol=CAT2025.08 | 52.3 | — | — | — | — | — | — | |
| SparseViT2026.04 | 51.65 | 93.86 | — | — | — | — | — | |
| SparseViTProtocol=CAT2025.08 | 51.3 | — | — | — | — | — | — | |
| EVPProtocol=Fine-tuned2026.04 | 50.97 | — | — | — | — | — | — | |
| MesorchProtocol=Fine-tuned2026.04 | 49.47 | — | — | — | — | — | — | |
| PSCC-NetProtocol=Fine-tuned2026.04 | 48.55 | — | — | — | — | — | — | |
| MVSSProtocol=CAT-Net protocol2025.09 | 48.2 | — | — | — | 47.7 | 49.5 | — | |
| FRAMESelection=Learned, Fusion=Learned2026.05 | 46.8 | — | — | — | — | — | 35.4 | |
| MVSS-Net2026.04 | 46.63 | 86.32 | — | — | — | — | — | |
| TruforProtocol=CAT-Net protocol2025.09 | 45.7 | — | — | — | 47.3 | 53.1 | — | |
| TruforProtocol=CAT2025.08 | 45.1 | — | — | — | — | — | — | |
| TruFor2026.04 | 45.06 | 94.24 | — | — | — | — | — | |
| PSCCProtocol=CAT-Net protocol2025.09 | 44.8 | — | — | — | 54.1 | 55.5 | — | |
| TruFor2026.05 | 44.7 | — | — | — | — | — | 33.8 | |
| RITAProtocol=MVSS protocol2025.09 | 44.3 | — | — | — | 51.4 | 51.5 | — | |
| IML-ViTProtocol=MVSS2025.08 | 43.8 | — | — | — | — | — | — | |
| RGB-NFine-tuned=true2022.03 | 43.7 | 81.7 | — | — | — | — | — | |
| CAT-Net2026.04 | 42.73 | 91.68 | — | — | — | — | — | |
| CAT-NetProtocol=CAT-Net protocol2025.09 | 42.7 | — | — | — | 47.8 | 53.3 | — | |
| TruForProtocol=MVSS protocol2025.09 | 41.9 | — | — | — | 42.1 | 46.4 | — | |
| TruforProtocol=MVSS2025.08 | 41.9 | — | — | — | — | — | — | |
| ACBGProtocol=Fine-tuned2026.04 | 41.03 | — | — | — | — | — | — | |
| PSCC-Net2026.04 | 40.34 | 88.38 | — | — | — | — | — | |
| CAT-Net2026.05 | 40.2 | — | — | — | — | — | 30.4 | |
| ManTraNet2026.05 | 37.8 | — | — | — | — | — | 28.1 | |
| XGB-Ensemble (pyIFD)Fusion=XGB-Ensemble2026.05 | 37.1 | — | — | — | — | — | 27.9 | |
| RF-Ensemble (pyIFD)Fusion=RF-Ensemble2026.05 | 35.6 | — | — | — | — | — | 26.8 | |
| Heuristic-K + uniformSelection=Heuristic, Fusion=Uniform2026.05 | 33.2 | — | — | — | — | — | 24.8 | |
| MesorchProtocol=MVSS protocol2025.09 | 32.6 | — | — | — | 34.6 | 40.2 | — | |
| Best single pyIFDSelection=Best single2026.05 | 31.8 | — | — | — | — | — | 23.7 | |
| CAT-NetProtocol=MVSS protocol2025.09 | 29.6 | — | — | — | 36.1 | 37.8 | — | |
| CAT-NetProtocol=MVSS2025.08 | 29.6 | — | — | — | — | — | — | |
| Random-K + uniformSelection=Random, Fusion=Uniform2026.05 | 29.4 | — | — | — | — | — | 21.5 | |
| ObjectFormerProtocol=MVSS2025.08 | 29.4 | — | — | — | — | — | — | |
| SegFormer w. ReViProtocol=Pre-trained2026.04 | 28.93 | — | — | — | — | — | — | |
| SparseViTProtocol=MVSS2025.08 | 28.7 | — | — | — | — | — | — | |
| Uniform-all pyIFDSelection=None (all), Fusion=Uniform2026.05 | 28.1 | — | — | — | — | — | 20.4 | |
| MesorchProtocol=MVSS2025.08 | 27.6 | — | — | — | — | — | — | |
| Swin Transformer w. ReViProtocol=Pre-trained2026.04 | 27.5 | — | — | — | — | — | — | |
| LDM w. ReViProtocol=Pre-trained2026.04 | 26.77 | — | — | — | — | — | — | |
| EVPProtocol=Pre-trained2026.04 | 26.67 | — | — | — | — | — | — | |
| MVSS-NetProtocol=MVSS protocol2025.09 | 25.9 | — | — | — | 29.3 | 32.7 | — | |
| MVSS-NetProtocol=MVSS2025.08 | 25.9 | — | — | — | — | — | — | |
| PSCC-NetProtocol=MVSS protocol2025.09 | 23.1 | — | — | — | 37.7 | 36.3 | — | |
| PSCC-NetProtocol=MVSS2025.08 | 23.1 | — | — | — | — | — | — | |
| NCL-IMLProtocol=MVSS2025.08 | 22.5 | — | — | — | — | — | — | |
| TinySAM w. ReViProtocol=Pre-trained2026.04 | 21.35 | — | — | — | — | — | — | |
| MMFusion2026.05 | 21.3 | — | — | — | — | — | 15.2 | |
| IML-ViTProtocol=Pre-trained2026.04 | 19.86 | — | — | — | — | — | — | |
| Sparse-ViTProtocol=Pre-trained2026.04 | 18.76 | — | — | — | — | — | — | |
| MesorchProtocol=Pre-trained2026.04 | 17.46 | — | — | — | — | — | — | |
| TruForProtocol=Pre-trained2026.04 | 14.91 | — | — | — | — | — | — | |
| PSCC-NetProtocol=Pre-trained2026.04 | 13.35 | — | — | — | — | — | — | |
| MPCProtocol=Pre-trained2026.04 | 9.49 | — | — | — | — | — | — | |
| Mantra-NetProtocol=MVSS2025.08 | 9 | — | — | — | — | — | — | |
| ACBGProtocol=Pre-trained2026.04 | 8.31 | — | — | — | — | — | — | |
| AdaCFA2022.12 | — | — | 21.5 | 18.3 | — | — | — | |
| ADQ2022.12 | — | — | 16.7 | 16.5 | — | — | — | |
| CAT-Net v22022.12 | — | — | 58.2 | 38.1 | — | — | — | |
| CR-CNN2022.12 | — | — | 48.7 | 39.1 | — | — | — | |
| EXIF-SC2022.12 | — | — | 33.2 | 16.4 | — | — | — | |
| H-LSTMFine-tuned=true2022.03 | — | 71.2 | — | — | — | — | — | |
| HiFi-NetEvaluation Protocol=Pre-trained, Pre-training Dataset=Same as PSCC [45]2023.03 | — | 93.2 | — | — | — | — | — | |
| HiFi-NetEvaluation Protocol=Pre-trained, Pre-training Dataset=HiFi-IFDL2023.03 | — | 92.4 | — | — | — | — | — | |
| IF-OSN2022.12 | — | — | 47.2 | 30.4 | — | — | — | |
| J-LSTMFine-tuned=true2022.03 | — | 61.4 | — | — | — | — | — | |
| ManTra-NetEvaluation Protocol=Pre-trained2023.03 | — | 81.9 | — | — | — | — | — | |
| ManTraNetPre-training Data Size=64K2022.03 | — | 81.9 | — | — | — | — | — | |
| ManTraNet2022.12 | — | — | 48.6 | 31.7 | — | — | — | |
| MVSS-Net2022.12 | — | — | 65.9 | 51.4 | — | — | — | |
| Noiseprintnormalization=between 0 and 12022.12 | — | — | 34.2 | 22.9 | — | — | — | |
| ObjectFormerPre-training Data Size=62K2022.03 | — | 92.8 | — | — | — | — | — | |
| ObjectFormerEvaluation Protocol=Pre-trained2023.03 | — | 92.8 | — | — | — | — | — | |
| PSCC-NetEvaluation Protocol=Pre-trained2023.03 | — | 84.7 | — | — | — | — | — | |
| PSCC-Net2022.12 | — | — | 61.5 | 47.3 | — | — | — | |
| PSCCNetPre-training Data Size=100K2022.03 | — | 84.7 | — | — | — | — | — | |
| RRU-Net2022.12 | — | — | 33.9 | 27.9 | — | — | — | |
| SPANPre-training Data Size=96K2022.03 | — | 92.2 | — | — | — | — | — |