Table 4. Detail of the layers and parameters defined in the standard CNN. The input of the network is an image of size 256x256x3 and the output are two scores of the two classes considered. In the parameters column K indicates the number of filters of the layer. After each convolution layer, batch normalization was applied.

Layer	Operation	Input size	Parameters	Layer	Operation	Input size	Parameters
1	Convolution	256x256	3x3, K = 32	9	Max-pooling	57x57	2x2
2	Convolution	254x254	3x3, K = 32	10	Convolution	28x28	3x3, K = 32
3	Max-pooling	252x252	2x2	11	Convolution	26x26	3x3, K = 32
4	Convolution	126x126	3x3, K = 32	12	Max-pooling	24x24	2x2
5	Convolution	124x124	3x3, K = 32	13	Convolution	12x12	3x3, K = 32
6	Max-pooling	122x122	2x2	14	Convolution	10x10	3x3, K = 32
7	Convolution	61x61	3x3, K = 32	15	Convolution	8x8	K = 128,Dropout, p = 0.5
8	Convolution	59x59	3x3, K = 32	16	Soft-Max	128x1	2 classes