Convolutional Neural Networks (CNNs) are at the heart of this deep learning revolution for improving the task of image classification. CNNs are specialized neural networks to handle image data. As a quick brush-up, CNNs help us infer shift and space invariant features through their shared weight architectures, and are basically a variant of feed forward networks. We have already covered the basics of CNNs in detail in Chapter 3, Understanding Deep Learning Architectures, and Chapter 5, Unleashing the Power of Transfer Learning. Before we move on, readers are encouraged to have a quick refresher for a better understanding. The following image showcases a typical CNN in action:

Neural networks arrived on the scene of image classification competitions as early as 2011. GPU-trained networks were starting to win competitions. It was in 2012, for the first time, when a deep CNN improved upon the previous best to 83% on the ImageNet image classification task that the world took notice. The results were amazing enough to catch global attention and help with the proliferation of use cases being solved using deep learning.