Table 4.

Similarity facets between proposed VIN and VGG-16.

Key	Similarity facet
A	Employing small convolution kernels with size of (3 × 3)
B	Employ small max pooling kernel with size of (2 × 2)
C	Each CB contains a few repetitions of conv layers followed by a max pooling layer
D	Fully connected layers are put at the end of the deep network
E	Channel number increase as it goes from input to the last conv layer, later decreases as to output.
F	Size of FMs shortens as it goes from input to output