Table 3. MobileNetv2 architecture.

Layer	Output Size	Kernel Size	Expansion Factor	Stride	Repeat
image	224×224×3			2	1
conv2d	112×112×32			2	1
bottleneck	112×112×16		1	1	1
bottleneck	56×56×24		6	2	2
bottleneck	28×28×32		6	2	3
bottleneck	14×14×64		6	2	4
bottleneck	14×14×96		6	1	3
bottleneck	7×7×160		6	2	3
bottleneck	7×7×320		6	1	1
conv2d	7×7×1280	1×1			1
avgpool	1×1×1280	7×7			1
fullyConnected	2
softmax	2
classification	2