New face profile scans 36M faces in one second-Mashable

April 4, 2012

Surveillance System Can Recognize a Face From 36 Million Others in One Second

Peter Pachal

Mashable: March 24, 2012


Face-recognition technology is rapidly evolving as evidenced by this new surveillance camera system. It can not only recognize specific faces, but is able to compare a single face to 36 million others in just one second.

The system, made by Hitachi Kokusai Electric and reported by DigInfo TV, was shown at a security trade show recently. It’s able to achieve its blazing speed by not wasting time on image processing — it takes visual data directly from the camera to compare the face in real time. The software also groups faces with similar features, so it’s able to narrow down the field very quickly.

When the system finds candidates that could be a match to the person being scanned, it immediately displays their thumbnails. The user can then review the archived footage and see if the person is, say, a repeat customer if it’s being used in a business. And the usefulness to law enforcement is pretty obvious.

SEE ALSO: Facebook Turns On Facial Recognition For Tagging By Default

As with any technology, there are limits. The software assumes people are looking right into the camera or facing no more than 30 degrees off center. And they’d better be close, too — if the face takes up anything smaller than a 40 x 40-pixel square, there’s just not enough face there to go on.

It’s impressive nonetheless, but it’s also a bit chilling to think that as soon as you look at a camera, whoever’s watching probably knows exactly who you are.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: