Skip to main content
. Author manuscript; available in PMC: 2022 Jul 27.
Published in final edited form as: IEEE J Biomed Health Inform. 2021 Jul 27;25(7):2398–2408. doi: 10.1109/JBHI.2021.3060997

TABLE I:

MobileNet architecture used as baseline classification model.

Layer Type Kernel Shape / Stride Input Size
Conv. 16 × 32 / 2 1 × 9000
Depthwise Conv. 16 × 64 / 1 32 × 4500
Depthwise Conv. 16 × 128 / 2 64 × 4500
Depthwise Conv. 16 × 128 / 1 128 × 2250
Depthwise Conv. 16 × 256 / 2 128 × 2250
Depthwise Conv. 16 × 256 / 1 256 × 1125
Depthwise Conv. 16 × 512 / 2 256 × 1125
Depthwise Conv. 16 × 512 / 1 512 × 563
Depthwise Conv. 16 × 512 / 2 512 × 563
Depthwise Conv. 16 × 512 / 1 512 × 282
Depthwise Conv. 16 × 512 / 2 512 × 282
Depthwise Conv. 16 × 512 / 1 512 × 141
Depthwise Conv. 16 × 1024 / 2 512 × 141
Depthwise Conv. 16 × 1024 / 1 1024 × 71
Average Pooling Pool 1 × 71 1024 × 71
Fully Connected 1024 × 4 1024 × 1