'Shazam for shoes' shows how AI could transform your style

Shoegazer, a prototype sneaker-spotting app, demonstrates the ways artificial intelligence could change how we shop.

Richard Trenholm Former Movie and TV Senior Editor
Richard Trenholm was CNET's film and TV editor, covering the big screen, small screen and streaming. A member of the Film Critic's Circle, he's covered technology and culture from London's tech scene to Europe's refugee camps to the Sundance film festival.
Expertise Films, TV, Movies, Television, Technology
Richard Trenholm
3 min read

You're in the shoe store trying on the new Air Yeezy Superstar Whatevers. To know if they're going to put your swag on fleek, you need more than a mirror on the floor and a sales lackey forcing a smile.

You're walking through downtown Tokyo and you see a kid wearing the trillest kicks you've ever seen. You need those shoes in your life, but who makes them?

In both these situations, artificial intelligence could save your sneaker-loving soul.

That's the promise of an AI-powered app called Shoegazer. It's like Shazam, the app that recognises songs, only for shoes. Shoezam, if you like.

No? Suit yourself.

I checked out Shoegazer in the Shoreditch offices of Happy Finish, a high-tech British marketing agency seeking investment to develop AI technology. Sneakers are just the hook. The prototype app aims to show brands and consumers the possibilities AI and machine learning have for discovering, exploring and selling all sorts of products.

Let's start with Shoegazer. Point your phone at a pair of trainers -- that's British for sneakers -- and the app uses image recognition and transfer learning to tell you the make and model. In a shop, the app could be linked to an interactive mirror that displays you wearing the shoes with perhaps some custom artwork as a background.

For brands, the goal is to sell you products by giving extra context about how they will fit your personal style, as well as, in this case, your feet. Sneakers are a good example for the potential of AI in marketing. For many, they're more than just items of clothing. When you choose your new daps, you need to know about the shoe's cachet or subcultural significance.

From Run-DMC's funky fresh Adidas Superstars to Kanye West's Yeezys, music and sneakers are tightly laced together. The app could take advantage of this relationship, delivering, for example, an appropriate music playlist. It could also serve up iconic photos and videos of stars wearing the kicks in question.

Brands want to form that connection with customers. The golden word is "engagement", hooking you in and starting a conversation rather than blaring an advert in your face.

To do all this, the app first has to recognise shoes, and that requires training. Fifty pictures of a particular sneaker are used to train Shoegazer to recognise it in future. Again, it doesn't have to be shoes: brands could input images of other clothing or products to train the app.

As the editorial team of sneaker blog Hypebeast told me, tech and trainers have frequently gone hand-in-hand, er...or foot. They point to souped-up sneakers like the Nike Air Max, Reebok Pump or the Nike self-lacing shoes inspired by "Back to the Future".

Enlarge Image

Fans wait to buy Nike Air Yeezy 2 shoes, which saw sneakerheads camp outside stores around the globe.

Target Presse Agentur/Getty Images

Sneaker brands are already engaged in conversations with customers too.

"Nike's relationship with disabled athletes helped to shape the Flyease system, which allows for easy entry into basketball sneakers", points out Hypebeast. Similarly, Adidas responded to fan mods that cut the plastic cage off UltraBOOST shoes by manufacturing an official version without the cage, the UltraBOOST Uncaged.

Happy Finish says brands could tap into that conversation to build an AI database, encouraging sneakerheads to add images of their kicks. In return, you'd be rewarded with points or offers or, I dunno, Pokémon or something -- whatever kids are into these days.

That collection of data could build a vast catalogue of what people are wearing and give the AI another potential: spotting trends. AI doesn't have the fashion nous of an Anna Wintour or Chiara Ferragni to look at something and say "People are going to flip out for that, dahhhling". But like Google's DeepMind AI mining health records to spot when people are at risk of getting ill, artificial intelligence can collect an infinitely greater data set than any human could process and figure out people are going in a certain direction.

Someday, Happy Finish says, AI could even tell you what to wear. By analysing current trends and comparing that data to the weather and your calendar appointments, your computer could tell you the exact outfit to look on point all day.

On that day, the shoe really will be on the other foot.

Watch this: Everything we know about Nike's new self-lacing shoe: The HyperAdapt 1.0