Practical Deep Learning for Cloud, Mobile, and Edge

Chapter 12. Not Hotdog on iOS with Core ML and Create ML

“I’m a rich,” said Jian-Yang, a newly minted millionaire in an interview with Bloomberg (Figure 12-1). What did he do? He created the Not Hotdog app (Figure 12-2) and made the world “a better place.”

Figure 12-1. Jian-Yang being interviewed by Bloomberg News after Periscope acquires his “Not Hotdog” technology (image source: From HBO’s Silicon Valley)

To the few of us who may be confused (including a third of the authors of this book), we are making a reference to HBO’s Silicon Valley, a show in which one of the characters is tasked with making SeeFood—the “Shazam for food.” It was meant to classify pictures of food and give recipes and nutritional information. Hilariously, the app ends up being good only for recognizing hot dogs. Anything else would be classified as “Not Hotdog.”

There are a few reasons we chose to reference this fictitious app. It’s very much a part of popular culture and something many people can easily relate to. It’s an exemplar: easy enough to build, yet powerful enough to see the magic of deep learning in a real-world application. It is also very trivially generalizable to recognize more than one class of items.

Figure 12-2. The Not Hotdog app in action (image source: Apple App Store listing for the Not ...