We certainly covered a lot of ground here! We talked about how to extract features from images, how convolutional neural networks work, and then we built a convolutional neural network to a fully-connected network architecture. Along the way, we picked up lots of new jargon and concepts, too!
Hopefully, after reading this chapter, you feel that these image-classification techniques—knowledge of which you may have once considered the province of sorcerers—is actually just a series of mathematical optimizations carried out for intuitive reasons! And hopefully this content can help move you forward in tackling an image-processing project that interests you!