Table 2.

Parameter details of ICNN

Layer number	Layer type	Input	Kernel size	Activation function	Output
1	Convolution	33 × 33 × 4	3 × 3	-	31 × 31 × 32
2	Activation	31 × 31 × 32	-	ReLU	31 × 31 × 32
3	Convolution	31 × 31 × 32	3 × 3		29 × 29 × 32
4	Activation	29 × 29 × 32	-	ReLU	29 × 29 × 32
5	Max-pooling	29 × 29 × 32	3 × 3		14 × 14 × 32
6	Convolution	14 × 14 × 32	3 × 3		12 × 12 × 64
7	Activation	12 × 12 × 64	-	ReLU	12 × 12 × 128
8	Convolution	12 × 12 × 128	3 × 3		10 × 10 × 64
9	Activation	10 × 10 × 64	-	ReLU	10 × 10 × 64
10	Max-pooling	10 × 10 × 64	2 × 2		5 × 5 × 64
11	convolution	5 × 5(64 + 22)	3 × 3		3 × 3 × 512
12	activation	3 × 3 × 512	-	ReLU	3 × 3 × 512
13	Convolution	3 × 3 × 512	3 × 3		1 × 1 × 12
14	Activation	1 × 1 × 12	-	ReLU	1 × 1 × 12
15	Fully connected	1 × 1 × 12	-	Softma ×	-
16	Output	-	-	-	2