Deepfake Detection on CIFAKE (test)
98.62AccuracyPhase4DFD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Phase4DFD2026.01 | 98.62 | 99.88 | |
| DenseNetModel=DenseNet, Parameters (M)=7.9, FLOPs (G)=5.6, Pre-training=ImageNet [26]2024.06 | 98 | 99 | |
| DenseNet2026.01 | 98 | 99 | |
| BNext-MModel=BNext-M, Backbone Condition=trainable, Parameters (M)=133, FLOPs (G)=3.39, Pre-training=ImageNet [26]2024.06 | 97.35 | 99.62 | |
| BNext-Mvariant=M2026.01 | 97.35 | 99.62 | |
| BNext-TModel=BNext-T, Backbone Condition=trainable, Parameters (M)=29.8, FLOPs (G)=0.89, Pre-training=ImageNet [26]2024.06 | 97.29 | 99.65 | |
| BNext-Tvariant=T2026.01 | 97.29 | 99.65 | |
| BNext-SModel=BNext-S, Backbone Condition=trainable, Parameters (M)=67.1, FLOPs (G)=1.91, Pre-training=ImageNet [26]2024.06 | 96.96 | 99.55 | |
| BNext-Svariant=S2026.01 | 96.96 | 99.55 | |
| VGGModel=VGG, Parameters (M)=133, FLOPs (G)=7.63, Pre-training=ImageNet [26]2024.06 | 96 | 99 | |
| VGGNet2026.01 | 96 | 99 | |
| ResNet-50Model=ResNet-50, Parameters (M)=25.6, FLOPs (G)=4.8, Pre-training=ImageNet [26]2024.06 | 95 | 99 | |
| ResNet2026.01 | 95 | 99 | |
| CNN2026.01 | 86 | 93 | |
| BNext-T with frozen backboneModel=BNext-T, Backbone Condition=frozen, Parameters (M)=29.8, FLOPs (G)=0.89, Pre-training=ImageNet [26]2024.06 | 83.89 | 91.7 | |
| BNext-M with frozen backboneModel=BNext-M, Backbone Condition=frozen, Parameters (M)=133, FLOPs (G)=3.39, Pre-training=ImageNet [26]2024.06 | 82.77 | 90.73 | |
| BNext-S with frozen backboneModel=BNext-S, Backbone Condition=frozen, Parameters (M)=67.1, FLOPs (G)=1.91, Pre-training=ImageNet [26]2024.06 | 80.71 | 89.25 |