Table 2.

Structure of the model.

Input Size	Layer	Output Size	Number of Feature Maps	Kernel Size	Stride	Activation
-	Image	200 × 300 × 3	-	-	-	-
200 × 300 × 3	Convolution	200 × 300 × 64	64	3 × 3	1	ReLU
200 × 300 × 64	Convolution	200 × 300 × 64	64	3 × 3	1	ReLU
200 × 300 × 64	Max pooling	100 × 150 × 64	64	-	2	-
100 × 150 × 64	Convolution	100 × 150 × 128	128	3 × 3	1	ReLU
100 × 150 × 128	Convolution	100 × 150 × 128	128	3 × 3	1	ReLU
100 × 150 × 128	Max pooling	50 × 75 × 128	128	-	2	-
50 × 75 × 128	Convolution	50 × 75 × 256	256	3 × 3	1	ReLU
50 × 75 × 256	Convolution	50 × 75 × 256	256	3 × 3	1	ReLU
50 × 75 × 256	Global max pooling	1 × 256	-	-	-	-
1 × 256	Fully connected	1 × 512	-	-	-	ReLU
1 × 512	Fully connected	1	-	-	-	Sigmoid