Deepfake Detection on CharadesDF
92.8AccuracyHuman ensemble
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Human ensembleJudge Type=Human, Aggregation Strategy=Ensemble2026.03 | 92.8 | 92.6 | 97.9 | 1 | |
| Hybrid ensembleJudge Type=Hybrid human-AI, Aggregation Strategy=Ensemble2026.03 | 92.4 | 92.4 | 97.9 | 0 | |
| Human (indiv.)Judge Type=Human, Aggregation Strategy=Individual2026.03 | 78.4 | 78.3 | 84.6 | 17.1 | |
| AI ensembleJudge Type=AI, Aggregation Strategy=Ensemble2026.03 | 60.7 | 68.3 | 69.7 | 0.9 | |
| AI detector (indiv.)Judge Type=AI, Aggregation Strategy=Individual2026.03 | 53.7 | 53.3 | 58.3 | 31.7 |