Table 1.

Comparsions of network structure and layer parameters of convolutional neural networks (CNNs).

	Features	Giusti [13]	Proposed Model 1	Proposed Model 2
0	3 × 101 × 101	Inputs	Inputs	Inputs
1	32 × 98 × 98	Conv	Conv	Conv
2	32 × 49 × 49	Pool	Pool	Pool
3	-	Tanh	ReLU	BN + PReLU
4	32 × 46 × 46	Conv	Conv	Conv
5	32 × 23 × 23	Pool	Pool	Pool
6	-	Tanh	ReLU	BN + PReLU
7	32 × 20 × 20	Conv	Conv	Conv
8	32 × 10 × 10	Pool	Pool	Pool
9	-	Tanh	ReLU	BN + PReLU
10	32 × 7 × 7	Conv	Conv	Conv
11	32 × 3 × 3	Pool	Pool	Pool
12	-	Tanh	ReLU	BN + PReLU
13	288→200	FC1	FC1	FC1
14	-	Tanh	ReLU	PReLU
15	200→K	FC2	FC2	FC2