Table 2.

The details of the architecture of the regression model.

Layer Name	Output Size	Layers Info	Layer Name	Output Size	Layers Info
$conv1.1$	$240\times 320$	$3\times 3,64,stride2$	$pool3$	$15\times 20$	$2\times 2maxpool,stride2$
$conv1.2$	$120\times 160$	$3\times 3,64,stride1$	$conv4.1$	$15\times 20$	$3\times 3,128,stride1$
$pool1$	$60\times 80$	$2\times 2maxpool,stride2$	$conv4.2$	$15\times 20$	$3\times 3,128,stride1$
$conv2.1$	$60\times 80$	$3\times 3,64,stride1$	$dropout$	$15\times 20$	0.5
$conv2.2$	$60\times 80$	$3\times 3,64,stride1$	$flattern$	38,400	−
$pool2$	$30\times 40$	$2\times 2maxpool,stride2$	$fc1$	1024	−
$conv3.1$	$30\times 40$	$3\times 3,128,stride1$	$fc2$	8	−
$conv3.2$	$30\times 40$	$3\times 3,128,stride1$