Deepfake Attribution on OW-DFA-40 Protocol-2
82.8All AccuracyCAL
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| CALKnowledge of number of novel forgery types (KU)=known2025.12 | 82.8 | 87.2 | 90 | 67.5 | 80.2 | 63.5 | 97.6 | |
| CDALKnowledge of number of novel forgery types (KU)=known2025.12 | 80.2 | 83.9 | 85.7 | 65.7 | 76.2 | 57.3 | 94.1 | |
| CAL (w/o KU)Knowledge of number of novel forgery types (KU)=unknown2025.12 | 80 | 86.1 | 88.7 | 63 | 78.2 | 59.1 | 97.5 | |
| CPLKnowledge of number of novel forgery types (KU)=known2025.12 | 79.6 | 83.7 | 86.3 | 63.6 | 75.3 | 55.3 | 95 | |
| PALGCDKnowledge of number of novel forgery types (KU)=known2025.12 | 77.6 | 80.6 | 79.2 | 65 | 75.4 | 56.9 | 92.3 | |
| LPSKnowledge of number of novel forgery types (KU)=known2025.12 | 75.1 | 79.2 | 78.5 | 59.9 | 73.7 | 53.2 | 92.8 | |
| SimGCDKnowledge of number of novel forgery types (KU)=known2025.12 | 74.3 | 81.3 | 68.4 | 62 | 76.1 | 55.8 | 86.4 | |
| OwMatchKnowledge of number of novel forgery types (KU)=known2025.12 | 72.6 | 80.1 | 64.2 | 61.2 | 74.7 | 54.4 | 84.1 | |
| LegoGCDKnowledge of number of novel forgery types (KU)=known2025.12 | 72.2 | 81 | 58.3 | 64.5 | 77.1 | 57.5 | 80.8 | |
| ProtoGCDKnowledge of number of novel forgery types (KU)=known2025.12 | 61.8 | 70 | 41.5 | 58.1 | 68.1 | 48.7 | 69.3 |