Consider the fragment of the convolutional architecture given below:

. Input image: 1 x x x y . Convolutional layer: 5 x5 , 2 x 3 number of input and output channels filter size stride . ReLU . MaxPooling: 3x3 ,3 x 3 region size stride . Convolutional layer: 4 -> 6, 4 x 4, 2 x 2 . ReLU . MaxPooling: 2 x 2, 2 x 2 . Flattening (3D to 1D): 6 x 12 x 8 - 576 number of feature maps x size of the feature map (12 x 8)