Introduction

In the previous chapters, we learned about traditional neural networks and a number of models, such as the perceptron. We learned how to train such models on structured data for regression or classification purposes. Now, we will learn how we can extend their application to the computer vision field.

Not so long ago, computers were perceived as computing engines that could only process well-defined and logical tasks. Humans, on the other hand, are more complex since we have five basic senses that help us see things, hear noises, feel things, taste foods, and smell odors. Computers were only calculators that could operate large volumes of logical operations, but they couldn't deal with complex data. Compared to the abilities of humans, computers had very clear limitations.

There were some rudimentary attempts to “give sight" to computers by processing and analyzing digital images. This field is called computer vision. But it was not until the advent...