Table 2.

Neural networks’ architectures.

Encoder ($\mathcal{E}$)
Layer	No. filters	Kernel size	Strides	Padding
$3\times \text{Convolutional block}$	16	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Max pooling}$	−	$(2\times 2)$	$(2\times 2)$	−
$3\times \text{Convolutional block}$	32	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Max Pooling}$	−	$(2\times 2)$	$(2\times 2)$	−
$3\times \text{Convolutional block}$	64	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Max Pooling}$	−	$(2\times 2)$	$(2\times 2)$	−
$3\times \text{Convolutional block}$	128	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Max Pooling}$	−	$(2\times 2)$	$(2\times 2)$	−
$3\times \text{Convolutional block}$	256	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Max pooling}$	−	$(2\times 2)$	$(2\times 2)$	−
Decoder ($\mathcal{D}$, ${\mathcal{D}}_{\mathit{aux}}$)
Layer	No. filters	Kernel size	Strides	Padding
$3\times \text{Convolutional block}$	256	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Conv2DTranspose}$	128	$(3\times 3)$	$(2\times 2)$	$\text{same}$
$\text{Batch normalization}$
$\text{Activation: ReLU}$
$3\times \text{Convolutional block}$	128	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Conv2DTranspose}$	64	$(3\times 3)$	$(2\times 2)$	$\text{valid}$
$\text{Batch normalization}$
$\text{Activation: ReLU}$
$3\times \text{Convolutional block}$	64	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Conv2DTranspose}$	32	$(3\times 3)$	$(2\times 2)$	$\text{same}$
$\text{Batch normalization}$
$\text{Activation: ReLU}$
$3\times \text{Convolutional block}$	32	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Conv2DTranspose}$	16	$(3\times 3)$	$(2\times 2)$	$\text{valid}$
$\text{Batch normalization}$
$\text{Activation: ReLU}$
$3\times \text{Convolutional block}$	16	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Conv2DTranspose}$	1	$(3\times 3)$	$(2\times 2)$	$\text{valid}$
$\text{Batch normalization}$
$\text{Activation: ReLU}$
$\text{Conv2D}$	1	$(3\times 3)$	$(1\times 1)$	$\text{same}$
$\text{Batch normalization}$
$\text{Activation: sigmoid}$