In this chapter, we have covered the intuition and the technical details of how CNNs work. we also had a look at how to implement a basic architecture of a CNN in TensorFlow.

In the next chapter we'll demonstrate more advanced architectures that could be used for detecting objects in one of the image datasets widely used by data scientists. We'll also see the beauty of CNNs and how they come to mimic human understanding of objects by first realizing the basic features of objects and then building up more advanced semantic features on them to come up with a classification for them. Although this process happens very quickly in our minds, it is what actually happens when we recognize objects.