Table 1 Layered architecture of the proposed model
Layer | Number of filters | Kernel Size | Strides | Output Shape | Number of parameters | Connected to |
|---|---|---|---|---|---|---|
Input | 175*175*1 | 0 | – | |||
2D Convolution layer | 30 | 5*5 | 1*1 | 175*175*30 | 780 | Input |
Batch Normalisation | 175*175*30 | 90 | 2D Convolution layer | |||
Depthwise 2D Convolution layer | 1*1 | 1*1 | 175*175*30 | 60 | Batch Normalisation | |
Batch Normalisation 1 | 175*175*30 | 90 | Depthwise 2D Convolution layer | |||
Depthwise 2D Convolution layer 1 | 1*1 | 1*1 | 175*175*30 | 60 | Batch Normalisation 1 | |
Batch Normalisation 2 | 175*175*30 | 90 | Depthwise 2D Convolution layer 1 | |||
Add | 175*175*30 | 0 | Batch Normalisation 1 | |||
Batch Normalisation 2 | ||||||
2D Convolution layer 1 | 25 | 3*3 | 1*1 | 175*175*25 | 6775 | Add |
Batch Normalisation 3 | 175*175*25 | 75 | 2D Convolution layer 1 | |||
2D Convolution layer 2 | 25 | 3*3 | 1*1 | 175*175*25 | 5650 | Batch Normalisation 3 |
Batch Normalisation 4 | 175*175*25 | 75 | 2D Convolution layer 2 | |||
2D Average Pooling | 2*2 | 2*2 | 87*87*25 | 0 | Batch Normalisation 4 | |
2D Convolution layer 3 | 55 | 3*3 | 1*1 | 87*87*25 | 12430 | 2D Average Pooling |
Batch Normalisation 5 | 87*87*25 | 165 | 2D Convolution layer 3 | |||
Depthwise 2D Convolution layer 2 | 1*1 | 1*1 | 87*87*25 | 110 | Batch Normalisation 5 | |
Batch Normalisation 6 | 87*87*25 | 165 | Depthwise 2D Convolution layer 2 | |||
Depthwise 2D Convolution layer 3 | 1*1 | 1*1 | 87*87*25 | 110 | Batch Normalisation 6 | |
Batch Normalisation 7 | 87*87*25 | 165 | Depthwise 2D Convolution layer 3 | |||
Add 1 | 87*87*25 | 0 | Batch Normalisation 5 | |||
Batch Normalisation 7 | ||||||
2D Convolution layer 4 | 55 | 3*3 | 1*1 | 87*87*25 | 27280 | Add 1 |
Batch Normalisation 8 | 87*87*25 | 165 | Batch Normalisation 5 | |||
2D Average Pooling 1 | 2*2 | 2*2 | 87*87*25 | 0 | Batch Normalisation 8 | |
2D Convolution layer 5 | 55 | 3*3 | 1*1 | 43*43*25 | 27280 | 2D Average Pooling 1 |
Batch Normalisation 9 | 43*43*25 | 165 | 2D Convolution layer 5 | |||
2D Convolution layer 6 | 25 | 1*1 | 1*1 | 43*43*25 | 1400 | Batch Normalisation 9 |
Batch Normalisation 10 | 43*43*25 | 75 | 2D Convolution layer 6 | |||
2D Global Average Pooling | 25 | 0 | Batch Normalisation 10 | |||
Dense | 2 | 2 | 52 | 2D Global Average Pooling |