Fig. 13.

Representation of the CNN architecture. The architecture is composed of 2 convolutional layers, each followed by a dropout step and maxpooling layer. Then, a flatten layer is added to flatten the input. The output layer consists of 2 neurons activated by the Softmax function