Exercise

# Rolling in the deep

You have been asked by the local police department to produce a Deep Learning model for license plate reading.

Your input data are images of digits, 28 pixels wide and 28 pixels tall, each with a label stating which of the 10 possible digits is present on the picture.

The `Sequential()`

model is already loaded, which you will use to build a Deep Neural Network using the following layers:

`Conv2D()`

- 2D convolutional layer`MaxPooling2D()`

- pooling layer`Flatten()`

- flattening layer`Dense()`

- fully connected layer

Instructions

**100 XP**

- Initialize the model and set a
**2D convolutional layer**with 64 filters of size 3x3 at the input. - Add a
`MaxPooling2D()`

layer, with default parameters, followed by a**flattening layer**, to reshape the signal from a 2-dimensional to a 1-dimensional format. - Add a fully connected
`Dense()`

layer with a`softmax`

activation function and`10`

neurons for 10 target classes present in our training set. - Compile the model.