Deepfake Detection on FF++ cross-domain benchmark leaderboard · SOTA2 Research

Deepfake Detection on FF++ cross-domain

1.01Gender FFPRSynergistic Fairness Optimization framework

Evaluation Results

Method
Synergistic Fairness Optimization framework2025.11	1.01	1.33	59.82	9	6.26	51.34	19.6	8.34	42.82	60.51
Fairadapter2025.11	7.19	4.74	58.84	33.8	16.94	51.05	73.72	29.96	45.53	59.64
Ours2025.11	8.49	13.02	59.89	10.52	6.05	54.18	39.7	22.29	45.75	60.25
Fairadapter2025.11	8.59	4.74	58.84	33.8	16.94	51.05	73.72	29.96	45.53	59.64
DAG-FDD2025.11	8.68	9.44	58.72	21.09	19.06	50.24	57.9	22.92	43.89	59.22
DAW-FDD2025.11	9.67	6.94	52.78	29.98	13.43	46.21	99.66	30.44	36.95	54.78
Ori2025.11	10.77	11.89	57.26	41.89	22.92	50.34	95.46	35.1	42.76	57.66
DAW-FDD2025.11	14.85	16.54	57.38	21.41	12.57	48.82	62.3	32.43	42.63	57.61
Ori2025.11	16.14	16.35	55.29	22.42	16.94	54.16	81.51	33.2	47.02	56.06
DAG-FDD2025.11	18	23.36	56.06	17.45	11.17	50.6	69.19	29.73	46.13	56.82