The generator that we are using here is a simple convolution autoencoder that is a combination of two parts—an encoder and a decoder.
In the encoder, we have the following:
- The first layer is a convolution 2D layer with 32 filters of a size of 3*3, followed by batch normalization, with activation as relu, followed by downsampling done with AveragePooling2D of size 2*2
- The second layer is a convolution 2D layer with 64 filters of a size of 3*3, followed by batch normalization, with activation as relu, followed by downsampling with AveragePooling2D of a size of 2*2
- The third layer or the final layer in this encoder part is again a convolution 2D layer with 128 filters of a size of 3*3, batch normalization, with activation ...
