Visualizing the outcome of feature learning

So far, we have learned about how CNNs help us classify images, even when the objects in the images have been translated. We have also learned that filters play a key role in learning the features of an image, which, in turn, help in classifying the image into the right class. However, we haven't mentioned what the filters learn that makes them powerful.

In this section, we will learn about what these filters learn that enables CNNs to classify an image correctly by classifying a dataset that contains images of X's and O's. We will also examine the fully connected layer (flatten layer) to understand what their activations look like. Let's take a look at what the filters learn:

The code for this section is available as Visualizing_the_features'_learning.ipynb in the Chapter04 folder of this book's GitHub repository - https://tinyurl.com/mcvp-packt.

1. Download the dataset:

!wget https://www.dropbox.com/s/5jh4hpuk2gcxaaq...