Table 4.

Detailed architecture for the CNN.

Layers	Type	Filter Size	Stride	Output Dimension	Activation	Mode
Input	1			(64,64,3)		Valid
Convolution	2	3 × 3	(1,1)	(64,64,8)	ReLU
Max-pooling	3	2 × 2		(32,32,8)
Convolution	4	3 × 3		(32,32,8)
Max-pooling	5	2 × 2		(16,16,8)
Dense	6			(10,1)
Dense	7			(2,1)	Softmax