Table 2.

Structure of Mobilnetv3+SE.

Input	Operator	Exp size	#Out
224 × 224 × 3	conv2d, 3 × 3	–	16
112 × 112 × 16	bneck, 3 × 3	16	16
56 × 56 × 16	bneck, 3 × 3	72	24
28 × 28 × 24	bneck, 3 × 3	88	24
28 × 28 × 24	bneck, 5 × 5	96	40
14 × 14 × 40	bneck, 5 × 5	240	40
14 × 14 × 40	bneck, 5 × 5	240	40
14 × 14 × 40	bneck, 5 × 5	120	48
14 × 14 × 48	bneck, 5 × 5	144	48
14 × 14 × 48	bneck, 5 × 5	288	96
7 × 7 × 96	bneck, 5 × 5	576	96
7 × 7 × 96	bneck,5 × 5	576	96
7 × 7 × 96	conv2d, 1 × 1	–	576
7 × 7 × 576	Pool, 7 × 7	–	–
1 × 1 × 576	conv2d, 1 × 1	–	1,024
1 × 1 × 1,024	AdaptiveAvgPool2d	–	–
1 × 1 × 1,024	conv2d, 1 × 1	–	64
1 × 1 × 64	BatchNorm2d	–	64
1 × 1 × 64	conv2d, 1 × 1	–	64
1 × 1 × 64	ReLU	–	64
1 × 1 × 64	conv2d, 1 × 1	–	15