Adopting transfer learning for neuroimaging: a comparative analysis with a custom 3D convolution neural network model

Table 2 Specification of used models in terms of number of trainable parameters, size of generated features and reported accuracy on ImageNet for TL models

Model	Parameters	Features	Accuracy on ImageNet [25]
InceptionV3	23,851,784	2,048	94.49%
VGG16	138,357,544	512	91.9%
ResNet50	25,636,712	18 x 25 x 2048	92.9%
3D Model	62,997,012	1,024	-

ISSN: 1472-6947