An Xceptional way of looking at CNN models

Since the dawn of BatchNormalization, data scientists both young and old have scoured the layers of neural networks searching for ways to improve their models' ability to identify the difference between a nose and chin. Or at least, this has been the focus of Convolutional Neural Networks (CNN) when image detection is performed on the human face. Using headshot images of people, can we as a society, create a model that can accurately tell us the different parts of our face the same way our brains do when we get up in the morning and look in a mirror? The answer to that question is a constantly evolving one but here we explore a number of different CNN architectures in an attempt to provide our own. So what is a CNN and how does it work?