Temporal Forgery Localization on LAV-DF
98.92mAP@0.5MG-RWKV
Evaluation Results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MG-RWKVFeature=Visual+Audio2026.07 | 98.92 | — | — | — | — | — | — | — | 92.45 | 91.64 | — | — | — | 94.81 | 38.47 | 93.41 | 93.41 | |
| UMMAFormerFeature=Visual+Audio2026.07 | 98.83 | — | — | — | — | — | — | — | 92.42 | 92.1 | — | — | — | 95.54 | 37.61 | 92.47 | 92.48 | |
| UMMAFormerFeatures=Visual+Audio, E2E=false2026.07 | 98.83 | — | — | — | — | — | — | 77.33 | 92.42 | 92.1 | — | — | — | 95.54 | 37.61 | 92.47 | 92.48 | |
| MFMSFeature=Visual+Audio2026.07 | 98.47 | — | — | — | — | — | — | — | 90.46 | 90.02 | — | — | — | 94.15 | 27.8 | 90.65 | 90.69 | |
| MFMSFeatures=Visual+Audio, E2E=false2026.07 | 98.47 | — | — | — | — | — | — | 73.47 | 90.46 | 90.02 | — | — | — | 94.15 | 27.8 | 90.65 | 90.69 | |
| EVASFeatures=Visual+Audio, E2E=true2026.07 | 98.41 | — | — | — | — | — | — | 94.55 | 97.67 | 97.59 | — | — | — | 96.61 | 88.63 | 97.73 | 97.81 | |
| MFMSSupervision=fully2026.01 | 97.31 | 98 | 97.91 | 97.78 | 97.63 | 96.69 | 95.79 | 97.3 | 98.94 | 98.86 | 98.62 | 95.61 | 98.01 | — | — | — | — | |
| UMMAFormerFeature=Visual2026.07 | 97.3 | — | — | — | — | — | — | — | 90.85 | 90.19 | — | — | — | 92.96 | 25.68 | 91.14 | 91.18 | |
| UMMAFormerFeatures=Visual, E2E=false2026.07 | 97.3 | — | — | — | — | — | — | 71.98 | 90.85 | 90.19 | — | — | — | 92.96 | 25.68 | 91.14 | 91.18 | |
| MG-RWKVFeature=Visual2026.07 | 96.73 | — | — | — | — | — | — | — | 91.18 | 90.14 | — | — | — | 92.36 | 26.6 | 92.17 | 92.17 | |
| UMMAFormerSupervision=fully2026.01 | 96.7 | 97.69 | 97.57 | 97.37 | 97.11 | 95.96 | 94.9 | 96.76 | 98.63 | 98.53 | 98.22 | 95.14 | 97.63 | — | — | — | — | |
| TriDetFeature=Visual+Audio2026.07 | 96.29 | — | — | — | — | — | — | — | 89.71 | 88.69 | — | — | — | 86.84 | 23.64 | 90.39 | 91 | |
| TriDetFeatures=Visual+Audio, E2E=false2026.07 | 96.29 | — | — | — | — | — | — | 68.92 | 89.71 | 88.69 | — | — | — | 86.84 | 23.64 | 90.39 | 91 | |
| ActionFormerSupervision=fully2026.01 | 96.28 | 97.97 | 97.69 | 97.27 | 96.78 | 95.51 | 94.61 | 96.59 | 99.17 | 99.02 | 98.41 | 95.95 | 98.14 | — | — | — | — | |
| EVAS (Lightweight)Features=Visual+Audio, E2E=true2026.07 | 96.25 | — | — | — | — | — | — | 89.93 | 95.92 | 95.71 | — | — | — | 92.4 | 81.13 | 96.12 | 96.28 | |
| DiMoDifFeatures=Visual+Audio, E2E=false2026.07 | 95.5 | — | — | — | — | — | — | 68 | 92.7 | 91.4 | — | — | — | 87.9 | 20.6 | 93.7 | 94.2 | |
| ActionFormerFeature=Visual2026.07 | 95.34 | — | — | — | — | — | — | — | 89.63 | 88.41 | — | — | — | 90.2 | 23.73 | 90.33 | 90.41 | |
| ActionFormerFeatures=Visual, E2E=false2026.07 | 95.34 | — | — | — | — | — | — | 69.76 | 89.63 | 88.41 | — | — | — | 90.2 | 23.73 | 90.33 | 90.41 | |
| TriDetSupervision=fully2026.01 | 93.19 | 94.99 | 94.74 | 94.35 | 93.83 | 92.2 | 90.67 | 93.42 | 97.12 | 96.93 | 96.31 | 93.68 | 96.01 | — | — | — | — | |
| ICS-AVFeature=Visual+Audio2026.07 | 87.4 | — | — | — | — | — | — | — | — | — | — | — | — | 66.8 | 5.72 | — | — | |
| BA-TFDFeature=Visual+Audio2026.07 | 76.9 | — | — | — | — | — | — | — | 64.08 | 66.9 | — | — | — | 38.5 | 0.25 | 60.77 | 58.42 | |
| BA-TFDFeatures=Visual+Audio, E2E=false2026.07 | 76.9 | — | — | — | — | — | — | 38.55 | 64.08 | 66.9 | — | — | — | 38.5 | 0.25 | 60.77 | 58.42 | |
| RT-DeepLocSupervision=weakly2026.01 | 67.43 | 90.14 | 86.61 | 81.06 | 75.29 | 59.26 | 50.31 | 72.87 | 84.18 | 84.18 | 84.18 | 83.61 | 84.03 | — | — | — | — | |
| AVFusionFeature=Visual+Audio2026.07 | 65.38 | — | — | — | — | — | — | — | 59.26 | 62.98 | — | — | — | 23.89 | 0.11 | 54.8 | 52.11 | |
| AVFusionFeatures=Visual+Audio, E2E=false2026.07 | 65.38 | — | — | — | — | — | — | 29.79 | 59.26 | 62.98 | — | — | — | 23.89 | 0.11 | 54.8 | 52.11 | |
| BA-TFDFeature=Visual2026.07 | 58.55 | — | — | — | — | — | — | — | 58.77 | 62.49 | — | — | — | 28.6 | 0.16 | 53.86 | 50.29 | |
| BA-TFDFeatures=Visual, E2E=false2026.07 | 58.55 | — | — | — | — | — | — | 29.1 | 58.77 | 62.49 | — | — | — | 28.6 | 0.16 | 53.86 | 50.29 | |
| LOCOSupervision=weakly2026.01 | 36.84 | 62.4 | 55.09 | 50.78 | 45.18 | 31.65 | 28.02 | 44.28 | 52.31 | 52.31 | 52.31 | 52.19 | 52.28 | — | — | — | — | |
| BMNFeature=Visual2026.07 | 24.01 | — | — | — | — | — | — | — | 41.24 | 53.26 | — | — | — | 7.61 | 0.07 | 31.6 | 26.93 | |
| BMNFeatures=Visual, E2E=false2026.07 | 24.01 | — | — | — | — | — | — | 10.56 | 41.24 | 53.26 | — | — | — | 7.61 | 0.07 | 31.6 | 26.93 | |
| AGTFeature=Visual2026.07 | 17.85 | — | — | — | — | — | — | — | 34.23 | 43.15 | — | — | — | 9.42 | 0.11 | 24.59 | 16.71 | |
| AGTFeatures=Visual, E2E=false2026.07 | 17.85 | — | — | — | — | — | — | 9.13 | 34.23 | 43.15 | — | — | — | 9.42 | 0.11 | 24.59 | 16.71 | |
| MDSFeature=Visual2026.07 | 12.78 | — | — | — | — | — | — | — | 36.71 | 37.88 | — | — | — | 1.62 | 0 | 34.39 | 32.15 | |
| MDSFeatures=Visual, E2E=false2026.07 | 12.78 | — | — | — | — | — | — | 4.8 | 36.71 | 37.88 | — | — | — | 1.62 | 0 | 34.39 | 32.15 | |
| GEM-TFLTraining Dataset=AV-Deepfake1M2026.03 | 12.4 | 30.5 | 23.9 | 20.4 | 16.8 | 9.89 | 5.78 | 17.1 | — | — | — | — | — | — | — | — | — | |
| BMN (I3D)Feature=Visual2026.07 | 10.56 | — | — | — | — | — | — | — | 44.39 | 48.49 | — | — | — | 1.66 | 0 | 37.13 | 31.55 | |
| BMN (I3D)Features=Visual, E2E=false2026.07 | 10.56 | — | — | — | — | — | — | 4.07 | 44.39 | 48.49 | — | — | — | 1.66 | 0 | 37.13 | 31.55 | |
| WMMTTraining Dataset=AV-Deepfake1M2026.03 | 10.2 | 21.7 | 18.5 | 15.7 | 13.1 | 7.39 | 4.45 | 13 | — | — | — | — | — | — | — | — | — | |
| PFormerTraining Dataset=AV-Deepfake1M2026.03 | 9.28 | 26.3 | 22.7 | 16.6 | 12.7 | 6.44 | 2.15 | 13.7 | — | — | — | — | — | — | — | — | — | |
| FuSTALSupervision=weakly2026.01 | 8.91 | 31.55 | 25.4 | 19.16 | 13.48 | 5.58 | 2.95 | 15.29 | 39.07 | 39.05 | 38.75 | 36.09 | 38.24 | — | — | — | — | |
| CoLASupervision=weakly2026.01 | 8.59 | 31.25 | 25.93 | 19.42 | 13.35 | 5.23 | 2.71 | 15.21 | 41.19 | 41.18 | 40.84 | 37.79 | 40.25 | — | — | — | — | |
| MDPTraining Dataset=AV-Deepfake1M2026.03 | 2.33 | 15.5 | 9.1 | 7.76 | 4.01 | 1.83 | 0.42 | 5.85 | — | — | — | — | — | — | — | — | — | |
| SALSupervision=weakly2026.01 | 0.58 | 12.72 | 3.86 | 1.78 | 1.04 | 0.26 | 0.09 | 2.9 | 15.56 | 15.53 | 15.47 | 14.55 | 15.28 | — | — | — | — | |
| CoLATraining Dataset=AV-Deepfake1M2026.03 | 0.01 | 1.35 | 0.13 | 0.03 | 0.01 | 0.01 | 0.01 | 0.22 | — | — | — | — | — | — | — | — | — | |
| LOCOTraining Dataset=AV-Deepfake1M2026.03 | 0 | 0.27 | 0.04 | 0.01 | 0 | 0 | 0 | 0.05 | — | — | — | — | — | — | — | — | — |