Skip to main content
. 2026 Feb 7;26:472. doi: 10.1186/s12903-026-07727-7

Table 3.

Pre-trained model configuration parameters

Model Total Parameters Trainable Parameters Frozen Layers Trainable Layers Feature Dimension (Hybrid)
VGG16 138.4 M 7.1 M Layers 1–15 (13 conv layers) Layers 16–23 (3 conv + FC) 25,088
ResNet50 25.6 M 7.1 M Layers 1-143 (3 residual stages) Layers 144–175 (2 stages + FC) 2,048
Xception 22.9 M 8.4 M Layers 1-115 (entry + 4 middle flows) Layers 116–134 (4 middle + exit flows) 2,048