AI-Generated Image Detection on DRCT-2M v1.4 (test)
99.97LDM Detection RateConv-B
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Conv-B2025.04 | 99.97 | 100 | 99.97 | 95.84 | 64.44 | 82 | 80.82 | 60.75 | 99.27 | 62.33 | 99.8 | 83.4 | 73.28 | 61.65 | 51.79 | 50.41 | 79.11 | — | |
| CNNSpotBackbone=ResNet-502025.04 | 99.87 | 99.91 | 99.9 | 97.55 | 66.25 | 86.55 | 86.15 | 72.42 | 98.26 | 61.72 | 97.96 | 85.89 | 82.84 | 60.93 | 51.41 | 50.28 | 81.12 | — | |
| F3Net2025.04 | 99.85 | 99.78 | 99.79 | 88.66 | 55.85 | 87.37 | 68.29 | 63.66 | 97.39 | 54.98 | 97.98 | 72.39 | 81.99 | 65.42 | 50.39 | 50.27 | 77.13 | — | |
| Ours/CLIPBackbone=CLIP2025.04 | 99.7 | 99.7 | 99.69 | 99.67 | 99.71 | 99.4 | 99.48 | 99.4 | 99.62 | 99.7 | 99.68 | 99.64 | 99.51 | 99.61 | 99.67 | 97.8 | 99.5 | — | |
| Ours/DINOv2Backbone=DINOv22025.04 | 99.55 | 99.55 | 99.55 | 99.54 | 99.55 | 94.7 | 99.53 | 99.23 | 99.31 | 99.55 | 99.54 | 99.55 | 99.39 | 99.48 | 99.55 | 97.42 | 99.06 | — | |
| GramNet2025.04 | 99.4 | 99.01 | 98.84 | 95.3 | 62.63 | 80.68 | 71.19 | 69.32 | 93.05 | 57.02 | 89.97 | 75.55 | 82.68 | 51.23 | 50.01 | 50.08 | 76.62 | — | |
| CLIP/RN50Backbone=ResNet-502025.04 | 99 | 99.99 | 99.96 | 94.61 | 62.08 | 91.43 | 83.57 | 64.4 | 98.97 | 57.43 | 99.74 | 80.69 | 82.03 | 65.83 | 50.67 | 50.47 | 80.05 | — | |
| UnivFD2025.04 | 98.3 | 96.22 | 96.33 | 93.83 | 91.01 | 93.91 | 86.38 | 85.92 | 90.44 | 88.99 | 90.41 | 81.06 | 89.06 | 51.96 | 51.03 | 50.46 | 83.46 | — | |
| DRCT2025.04 | 94.45 | 94.35 | 94.24 | 95.05 | 95.61 | 95.38 | 94.81 | 94.48 | 91.66 | 95.54 | 93.86 | 93.48 | 93.54 | 84.34 | 83.2 | 67.61 | 91.35 | — | |
| De-fake2025.04 | 92.1 | 99.53 | 99.51 | 89.65 | 64.02 | 69.24 | 92 | 93.93 | 99.13 | 70.89 | 58.98 | 62.34 | 66.66 | 50.12 | 50.16 | 50 | 75.52 | — | |
| OmniAID-MirageTraining Dataset=Mirage-Train2025.11 | 16.54 | 0.88 | 0.92 | 4.58 | 13.74 | 3.12 | 33.1 | 54.58 | 2.06 | 0.56 | 9.4 | 52.76 | 29.38 | 0.02 | 0.14 | — | 13.95 | 1.48 | |
| De-fakeTraining Dataset=SD v1.42025.11 | 15.46 | 0.6 | 0.64 | 20.36 | 71.62 | 61.18 | 15.66 | 11.8 | 1.4 | 57.88 | 81.7 | 74.98 | 66.34 | 99.42 | 99.34 | — | 48.63 | 99.66 | |
| DIRETraining Dataset=SD v1.42025.11 | 3.56 | 0.06 | 0.02 | 63.62 | 92.26 | 56.08 | 82.2 | 91.24 | 0.38 | 80.48 | 0.64 | 71.54 | 81.68 | 95.96 | 99.86 | — | 57.47 | 100 | |
| UnivFDTraining Dataset=SD v1.42025.11 | 2.54 | 6.7 | 6.48 | 11.48 | 17.12 | 11.32 | 26.38 | 27.3 | 18.26 | 21.16 | 18.32 | 37.02 | 21.02 | 95.24 | 97.08 | — | 32.23 | 98.22 | |
| GramNetTraining Dataset=SD v1.42025.11 | 0.5 | 1.28 | 1.62 | 8.7 | 74.04 | 37.94 | 56.92 | 60.66 | 13.2 | 85.26 | 19.36 | 48.2 | 33.94 | 96.84 | 99.28 | — | 46.06 | 99.14 | |
| CNNSpotTraining Dataset=SD v1.42025.11 | 0.16 | 0.08 | 0.1 | 4.8 | 67.4 | 26.8 | 27.6 | 55.06 | 3.38 | 76.46 | 3.98 | 28.12 | 34.22 | 78.04 | 97.08 | — | 37.66 | 99.34 | |
| F3NetTraining Dataset=SD v1.42025.11 | 0.12 | 0.26 | 0.24 | 22.5 | 88.12 | 25.08 | 63.24 | 72.5 | 5.04 | 89.86 | 3.86 | 55.04 | 35.84 | 68.98 | 99.04 | — | 45.56 | 99.28 | |
| Conv-BTraining Dataset=SD v1.42025.11 | 0.06 | 0 | 0.06 | 8.32 | 71.12 | 36 | 38.36 | 78.5 | 1.46 | 75.34 | 0.4 | 33.2 | 53.44 | 76.7 | 96.42 | — | 41.79 | 99.18 | |
| CLIP/RN50Training Dataset=SD v1.42025.11 | 0 | 0 | 0.06 | 10.76 | 75.82 | 17.12 | 32.84 | 71.18 | 2.04 | 85.12 | 0.5 | 38.6 | 35.92 | 68.32 | 98.64 | — | 39.75 | 99.04 | |
| DRCTTraining Dataset=SD v1.42025.11 | 0 | 0.02 | 0.02 | 7.18 | 32.08 | 28.56 | 16.06 | 59.74 | 0.5 | 42.3 | 0.02 | 0.98 | 37.4 | 0.02 | 9.02 | — | 17.68 | 49.04 |