Tess Nichol is an NZME. News Service reporter.

Don't know what building that is? There's an app for that

Photo / Getty Images
Photo / Getty Images

Ever looked in wonder at a landmark and wondered what on Earth it might be?

Auckland tech wizards have created an image recognition app that might be able to help.

The "WTF Is That" web robot is the brainchild of Spark 64, a North Shore tech company.

People can send an image of what they want identified to the bot's Facebook page, using the website's chat function Messenger.

Designer Ming Cheuk said the bot used a combination of different image recognition services available from companies including Google and Microsoft.

It launched two weeks ago and has already identified more than 16,000 images sent in from thousands of users around the world.

Cheuk, who created the bot with colleagues Daniel Xu and Richard McLean, said the bot was in early stages and had some limitations.

The Statue of Liberty, for example, is easy to recognise, but when sent a picture of Auckland's Sky Tower it said: "Your image is a grey high-rise building tourist spot".

"Machines are good at recognising some thing but other things are difficult for a machine to understand," said Cheuk.

The bot would improve with tweaks over time - and could be upgraded so people can interact.

"Once we start building our more detailed responses, it will add more of a description of what it sees as far as it knows.

"It might even give you a link to the Wikipedia page [for the object pictured]."

Cheuk saw plenty of potential to turn it into something more than just a fun novelty.

One avenue the trio are exploring was to work with fashion companies to teach the bot to recognise items of clothing within an image and direct users to the brand's website to buy the product online.

- Herald on Sunday

Get the news delivered straight to your inbox

Receive the day’s news, sport and entertainment in our daily email newsletter

SIGN UP NOW

© Copyright 2016, NZME. Publishing Limited

Assembled by: (static) on production bpcf04 at 07 Dec 2016 22:28:09 Processing Time: 409ms