Table 1 Highest model accuracy metrics on the test set for two-class classification over all k-folds. Note: Highest values for each metric highlighted in bold
From: Comparison of deep learning models to detect crossbites on 2D intraoral photographs
Model | Accuracy (in %) | Specificity (in %) | Precision (in %) | Recall (sensitivity) (in %) | F1-score (in %) | Cohen’s Kappa (in %) |
|---|---|---|---|---|---|---|
ResNet18 | 97.14 | 100 | 98.00 | 95.45 | 96.60 | 93.20 |
ResNet50 | 91.43 | 90.91 | 90.48 | 91.61 | 90.96 | 81.93 |
MobileNet | 98.55 | 97.62 | 98.21 | 98.81 | 98.49 | 96.98 |
Xception | 98.57 | 100 | 98.94 | 97.92 | 98.40 | 96.80 |
DenseNet | 97.10 | 95.24 | 96.55 | 97.52 | 96.99 | 93.99 |
EfficientNet | 97.10 | 95.65 | 96.00 | 97.83 | 96.81 | 93.62 |