In this chapter, we've demonstrated how to design and implement a CNN to detect and recognize logos in photos. These photos were obtained from social media using the Twitter API and some search keywords. As photos are found, and logos were detected, records of the detections were saved to a CSV file for later processing or viewing.
Along the way, we looked at the origin of the term deep learning and discussed the various components that led to a revolution in ML in the last few years. We showed how a convergence of technologies and sociological factors helped this revolution. We've also seen multiple demonstrations of how far we have come in such a short amount of time – with very little code, a set of training examples (made available to researchers for free), and a GPU, we can create our deep neural network with ease.
In the next chapter, we'll show how to use statistics and other techniques to discover trends and recognize anomalies such as a dramatic increase or decrease in the...