Fig. 8.

Comparative performance analysis of fine-tuned pre-trained CNN models for dental condition classification. a Performance metrics comparison across three pre-trained models showing accuracy, precision, recall, and F1-score with error bars representing standard deviation from 5-fold cross-validation. b Confusion matrices for each pre-trained model displaying classification results across four dental conditions (fillings, cavity, implant, and impacted tooth), with numerical values indicating the distribution of predicted versus true labels. c Training and validation loss curves for VGG16, Xception, and ResNet50 models across training epochs, demonstrating convergence behavior and generalization characteristics for each architecture