Table 1.

Details of the visual geometry group (VGG)-based networks configurations with respect to number and type of layers (depth), filter size, number of filters, and the output size after each operation. The input image size and output size shown in the table are for the mathematical analysis of images (AMI) and AMI cropped (AMIC) datasets. For the West Pomeranian University of Technology (WPUT) dataset, the models use an input image size of $170\times 220$ to preserve the aspect ratio and the output after each operation is obtained similar to AMI and AMIC.

Block	Model				Filter Size	Output Size
	VGG-11	VGG-13	VGG-16	VGG-19
Input	Input Image ($133\times 190$ RGB)				-	-
Block 1	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (64)	$133\times 190$
		Convolution	Convolution	Convolution	$3\times 3$ (64)	$133\times 190$
	Max-Pooling				-	$66\times 95$
Block 2	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (128)	$66\times 95$
		Convolution	Convolution	Convolution	$3\times 3$ (128)	$66\times 95$
	Max-Pooling				-	$33\times 47$
Block 3	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (256)	$33\times 47$
	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (256)	$33\times 47$
			Convolution	Convolution	$3\times 3$ (256)	$33\times 47$
				Convolution	$3\times 3$ (256)	$33\times 47$
	Max-Pooling				-	$16\times 23$
Block 4	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (512)	$16\times 23$
	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (512)	$16\times 23$
			Convolution	Convolution	$3\times 3$ (512)	$16\times 23$
				Convolution	$3\times 3$ (512)	$16\times 23$
	Max-Pooling				-	$8\times 11$
Block 5	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (512)	$8\times 11$
	Convolution	Convolution	Convolution	Convolution	$3\times 3$ (512)	$8\times 11$
			Convolution	Convolution	$3\times 3$ (512)	$8\times 11$
				Convolution	$3\times 3$ (512)	$8\times 11$
	Adaptive Average Pooling				-	$5\times 5$
Fully Connected	Fully Connected, 2048 Neurons
	Dropout Chance $50\%$
	Fully Connected, 2048 Neurons
	Dropout Chance $50\%$
	Fully Connected, 100 Neurons
	Log-Soft-Max