Skip to main content
. 2018 Jul 23;32(1):105–115. doi: 10.1007/s10278-018-0107-6

Table 2.

Architecture of the proposed DCNN model

Layer Type Maps Map size Kernel size Stride Padding Activation
out Fully connected 5 Softmax
F7 Fully connected 100 ReLU
F6 Fully connected 5670 ReLU
C5 Convolution 70 9 × 9 3 × 3 1 1 ReLU
C4 Convolution 60 9 × 9 3 × 3 2 1 ReLU
C3 Convolution 50 17 × 17 3 × 3 1 1 ReLU
C2 Convolution 70 17 × 17 3 × 3 2 1 ReLU
C1 Convolution 50 33 × 33 3 × 3 1 1 ReLU
In Input 4 33 × 33