Table 3.

FCNs architecture.

Layer	Kernel Size. Feature Map	Stride	Output Shape
Input Image	-	-	256 × 256 × 1
Convolution Layer 1	28 × 28 × 256	2	3 × 3
Pooling Layer 1	2 × 2		14 × 14 × 256
Convolution Layer 2	14 × 14× 256	2	3 × 3
Pooling layer 1	2 × 2		7 × 7 × 256
Convolution Layer 3	7 × 7 × 7 × 256 × 7 × 256	2	-
Deconvolution	2 × 2		14 × 14 × 256
Convolution Layer 4	14 × 14 × 256	2	3 × 3
Deconvolution	2 × 2		28 × 28 × 256
Convolution Layer 5	28 × 28 × 256		3 × 3
Convolution	28 × 28 × C		28 × 28 × 1
Output Layer	-	-	1