Table 1.
Layers | Input | Kernel number | Kernel size | Stride | Pad | Output | ||||
---|---|---|---|---|---|---|---|---|---|---|
W | H | D | W | H | D | |||||
Input | 28 | 28 | 1 | |||||||
conv1 | 28 | 28 | 1 | 6 | 5 | 1 | 0 | 24 | 24 | 6 |
cccp1 | 24 | 24 | 6 | 6 | 1 | 1 | 0 | 24 | 24 | 6 |
cccp2 | 24 | 24 | 6 | 6 | 1 | 1 | 0 | 24 | 24 | 6 |
maxpool1 | 24 | 24 | 6 | 2 | 2 | 0 | 12 | 12 | 6 | |
conv2 | 12 | 12 | 6 | 12 | 5 | 1 | 0 | 8 | 8 | 12 |
cccp3 | 8 | 8 | 12 | 12 | 1 | 1 | 0 | 8 | 8 | 12 |
cccp4 | 8 | 8 | 12 | 12 | 1 | 1 | 0 | 8 | 8 | 12 |
avgpool2 | 8 | 8 | 12 | 2 | 2 | 0 | 4 | 4 | 12 | |
conv3 | 4 | 4 | 24 | 24 | 4 | 1 | 0 | 1 | 1 | 24 |
cccp5 | 1 | 1 | 24 | 24 | 1 | 1 | 0 | 1 | 1 | 24 |
cccp6 | 1 | 1 | 24 | 2 | 1 | 1 | 0 | 1 | 1 | 2 |
Avgpool3 | 1 | 1 | 2 | 2 | 2 | 0 | 1 | 1 | 2 | |
Softmax-loss | 1 | 1 | 2 | 1 | 1 | 2 |