Table 2.

Architecture of the MobileNet with transfer learning.

	Layer	Layer Type		Stride	Kernel Size	Input Size	N°Parameters
MobileNet Base Model	1	Conv. 2D		s2	$3\times 3\times 3\times 16$	$128\times 128\times 3$	496
	2	Conv. dw		s1	$3\times 3\times 16$	$64\times 64\times 16$	208
	3	Conv. pw		s1	$1\times 1\times 16\times 32$	$64\times 64\times 16$	640
	4	Conv. dw		s2	$3\times 3\times 32$	$64\times 64\times 32$	416
	5	Conv. pw		s1	$1\times 1\times 32\times 64$	$32\times 32\times 32$	2304
	6	Conv. dw		s1	$3\times 3\times 64$	$32\times 32\times 64$	832
	7	Conv. pw		s1	$1\times 1\times 64\times 64$	$32\times 32\times 64$	4352
	8	Conv. dw		s2	$3\times 3\times 64$	$32\times 32\times 64$	832
	9	Conv. pw		s1	$1\times 1\times 64\times 128$	$16\times 16\times 64$	8704
	10	Conv. dw		s1	$3\times 3\times 128$	$16\times 16\times 128$	1664
	11	Conv. pw		s1	$1\times 1\times 128\times 128$	$16\times 16\times 128$	16,896
	12	Conv. dw		s2	$3\times 3\times 128$	$16\times 16\times 128$	1664
	13	Conv. pw		s1	$1\times 1\times 128\times 256$	$8\times 8\times 128$	33,792
	14–23	$5\times$	Conv. dw	s1	$3\times 3\times 256$	$8\times 8\times 256$	$5\times 3328$
			Conv. pw	s1	$1\times 1\times 256\times 256$	$8\times 8\times 256$	$5\times \mathrm{66,560}$
	24	Conv. dw		s2	$3\times 3\times 256$	$8\times 8\times 256$	3328
	25	Conv. pw		s1	$1\times 1\times 256\times 512$	$4\times 4\times 256$	133,120
	26	Conv. dw		s1	$3\times 3\times 512$	$4\times 4\times 512$	6656
	27	Conv. pw		s1	$1\times 1\times 512\times 512$	$4\times 4\times 512$	264,192
Dense	–	Global Avg. Pool		s1	Pool $4\times 4$	$4\times 4\times 512$	-
	28	FC		–	–	512	262,656
	–	Softmax		–	Output	2	1026
Total Parameters: 1,093,218
Trainable Parameters: 263,682