acm-header
Sign In

Communications of the ACM

ACM TechNews

Computers Are Learning to See the World Like We Do


View as: Print Mobile App Share:
A human eye.

For the past year, Flickr has been training neural networks to determine if a given photo contains one of 1,000 known objects, ranging from a cat to a sunset.

Credit: mysticmundane.blogspot.com

Object recognition is challenging for computers, but artist Randall Munroe has launched a website called Park or Bird on Flickr intended to determine whether a photo was taken in a national park or shows a bird. Users can drag a photograph into the page, and it will make an educated guess.

Gerry Pesavento of Yahoo, which owns Flickr, says the site is "showing that image intelligence is happening very quickly."

For the past year, Flickr has been training neural networks to determine if a given picture has one of 1,000 different objects in it, ranging from a cat to a sunset. If Flickr can resolve this problem, it will significantly improve the search function for its billions of photos, according to Yahoo. Users potentially could search for images of any item even if the photographer has not tagged it.

Object-recognition technology also could eventually be used in autonomous vehicles or help visually impaired individuals identify people and street signs. "The sort of algorithms that we're using at Flickr right now are exactly the sort of algorithms that are going to be helping robots see and navigate visually," says Yahoo's Simon Osindero.

From New Scientist
View Full Article

 

Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account