Table 6 Accuracy of the reference deep learning models for each PAIs technique and bona fide samples on the k-fold cross-validation and unbalanced data partition.
Model | Bona Fide | Inpainting | Crop & Replace |
|---|---|---|---|
EfficientNet | 99.9 ± 0.10 | 99.4 ± 0.78 | 100.0 ± 0.00 |
ResNet | 99.9 ± 0.18 | 99.5 ± 0.66 | 100.0 ± 0.00 |
VIT | 99.3 ± 0.73 | 86.2 ± 8.06 | 93.8 ± 9.65 |
TransFG | 99.9 ± 0.15 | 99.4 ± 1.64 | 99.6 ± 0.61 |
CoAARC | 99.1 ± 1.15 | 98.3 ± 2.89 | 100.0 ± 0.00 |
