Hands-On Python Deep Learning for the Web

Reshaping the arrays for processing with Keras

The current shape of the image arrays are not Keras-friendly. We must convert the image arrays into a shape of (60000, 28, 28, 1) and (10000, 28, 28, 1), respectively.

To do so, we use the following lines of code:

x_train = train_images.reshape(train_images.shape[0], 28, 28, 1)x_test = test_images.reshape(test_images.shape[0], 28, 28, 1)

Now, if we observe the shape of x_train, we get an output as follows:

(60000, 28, 28, 1)

We have no changes to make in the labels arrays and so we directly assign them to y_train and y_test:

y_train = train_labelsy_test = test_labels

Next, we will create a neural network using Keras.