Research Article
Recognition of Gurmukhi Handwritten City Names Using Deep Learning and Cloud Computing
Table 5
Layered architecture for each layer.
| Layers | Input image size | Filter size | No of filters | Activation function | Output without padding | Parameters |
| Input image | 64 32 3 | | | | | | Convolution | 64 32 3 | 3 3 | 32 | ReLU | 64 32 32 | 896 | Max Pooling | 64 32 32 | Pool size (2 2) | — | — | 32 16 32 | 0 | Convolution | 32 16 32 | 3 3 | 64 | ReLU | 32 16 64 | 18496 | Max Pooling | 32 16 64 | Pool size (2 2) | — | — | 16 8 64 | 0 | Max Pooling | 16 8 64 | Pool size (2 2) | — | — | 8 4 64 | 0 | Flatten | 8 4 64 | — | — | — | 1024 | — | Dense | 1024 | 120 | — | ReLU | 120 | 123000 | Dense | 120 | — | — | SoftMax | 10 | 1210 |
|
|