Visual schematic for the full neural network. White regions are the CNN, light shaded regions are the FCN and the dark shaded regions are the Top Network. The input to the CNN is a single band, 128 × 128 pixel image while the input to the FCN is the 17 morphological parameters. Both of these inputs are on the left of their corresponding part of the network. The output from the full network is on the right as a two neuron, binary classification. The sizes of the kernels for the CNN (red) and fully connected layers for all parts of the network are shown. The blue line between layers represent the batch normalisation, ReLU activation and dropout that is applied between layers. The CNN has 67 652 128 trainable parameters, the FCN has 19 328 trainable parameters and the Top Network has 66 818 trainable parameters. The full network has a total of 67 738 274 trainable parameters.