Skip to main content
. 2021 Dec 1;21(23):8025. doi: 10.3390/s21238025

Table 1.

Architecture of the deep convolutional neural network.

Layer Channels In Channels Out Kernel Size Padding Stride
Conv2d 3 8 3 × 3 1 × 1 1 × 1
MaxPool2d 8 8 3 × 3 0 × 0 3 × 3
Conv2d 8 16 3 × 3 1 × 1 1 × 1
MaxPool2d 16 16 3 × 3 0 × 0 3 × 3
Conv2d 16 32 3 × 3 1 × 1 1 × 1
MaxPool2d 32 32 3 × 3 0 × 0 3 × 3
Conv2d 32 64 3 × 3 1 × 1 1 × 1
MaxPool2d 64 64 3 × 3 0 × 0 3 × 3
Conv2d 64 128 3 × 3 1 × 1 1 × 1
MaxPool2d 128 128 3 × 3 0 × 0 3 × 3
Conv2d 128 2 1 × 1 0 × 0 1 × 1