Table 2.

Specification of DCSCNN Architecture.

Layers	Output Size	DCSCNN
Input	256 × 256	-
Convolution (1)	128 × 128	5 × 5 conv, stride 2, filter 64
Dense Block (1)	128 × 128	[3 × 3 conv] × 2, filter 128, 128
Squeeze Block (1)	128 × 128	$\left[\begin{array}{c}1\times 1 conv\\ 1\times 1 conv\\ 3\times 3 conv\end{array}\right]$, filter 64, 128, 128
Max Pooling (1)	128 × 128	2 × 2, stride 2
Dense Block (2)	64 × 64	[3 × 3 conv] × 2, filter 256, 512
Squeeze Block (2)	64 × 64	$\left[\begin{array}{c}1\times 1 conv\\ 1\times 1 conv\\ 3\times 3 conv\end{array}\right]$, filter 128, 256, 256
Max Pooling (2)	32 × 32	2 × 2, stride 2
Dense Block (3)	32 × 32	[3 × 3 conv] × 2, filter 512, 256
Squeeze Block (3)	32 × 32	$\left[\begin{array}{c}1\times 1 conv\\ 1\times 1 conv\\ 3\times 3 conv\end{array}\right]$, filter 256, 512, 512
Max Pooling (3)	16 × 16	2 × 2, stride 2
Dense Block (4)	16 × 16	[3 × 3 conv] × 2, filter 128, 128
Squeeze Block (4)	16 × 16	$\left[\begin{array}{c}1\times 1 conv\\ 1\times 1 conv\\ 3\times 3 conv\end{array}\right]$, filter 64, 128, 128
Convolution (2)	16 × 16	1 × 1 conv, filter 64
Classification Block	16 × 16	FC (32)
		FC (16)
		FC (2 & 4), SoftMax