Table 3.
Architecture description of the CNN-2 : convolutional neural network-1 section of the proposed system.
Layer | #Kernel | Kernel size | #Parameter | Output size |
---|---|---|---|---|
Input | 0 | 0 | 0 | H × W × 1 |
C 1 1 | 32 | 3 × 3 | 320 | H × W × 32 |
C 1 2 | 32 | 3 × 3 | 9248 | H × W × 32 |
C 2 1 | 32 | 3 × 3 | 18496 | H/2 × W/2 × 32 |
C 2 2 | 48 | 3 × 3 | 27696 | H/2 × W/2 × 48 |
C 2 3 | 32 | 3 × 3 | 27712 | H/2 × W/2 × 32 |
C 3 1 | 64 | 3 × 3 | 73856 | H/4 × W/4 × 64 |
C 3 2 | 48 | 3 × 3 | 73792 | H/4 × W/4 × 48 |
C 3 3 | 64 | 3 × 3 | 36928 | H/4 × W/4 × 64 |
Total number of parameters in CNN-2: 268048 |