Friday, June 19, 2009
New Facebook Application Identifies People's Identities, Even "In the Wild," Using Constantly Evolving Artificial Intellegence. Scary Stuff.
This is probably going to be the number one facebook app. in about 2.1 seconds.
OK, call me paranoid, but even if this technology is meant for "fun" and "social" applications, it still has the ominous ability to identify people very quickly, and it will probably be able to pick people out "in the wild" in a very short time because it uses artificial intelligence.
So, let's say that someone takes my picture at a pro-Israel rally, feeds it into "friend finder," and finds out who I am, cross-references my address, and shows up at my house to do a little graffiti work, or, worse, brings a gun.
What if they cross reference to find my kids' facebook pages and their friends facebook pages and target a whole community by just seeing one of us?
My kids are always teasing me, telling me I am way too worried about "random" things in the world like security applications, nanotechnology, and arab terror. They can tease all they want.
I've lived a few years longer, and I know the world is not the safe place I have raised them to believe it is.
This sounds like the best-friend application of freedom-fighter and terrorist alike.
Sure, it would allow someone who was groped on the subway to identify the attacker, but it could also allow an attacker to identify a victim!
I think the world is just getting a bit too small for me.
Onstar in my car was scary enough--this stuff is terrifying!
Who's that on Facebook? Ask Friend Finder
By David Shamah
June 16, 2009
It can be tricky trying to figure out who everyone is in your Facebook photo albums, but now an Israeli company has come up with a solution - a unique face recognition analysis technology that can analyze and identify people in your online photos, and even connect you with them on the web.
Developed by Face.com, the Friend Finder application is designed for the millions of Facebook users who paste their pictures on the social networking site. Since the site was first founded, users have uploaded more than 15 billion pictures on the website, and a massive 60 percent of all photos on the web can be found on Facebook pages.
"In the first month of our alpha test, we scanned some 400 million photos, identifying about 700,000 people, with users confirming the identities of about 150,000 people," says Gil Hirsch Face.com's CEO and co-founder. "Based on the number of people we had in the test, that works out to an astounding average of 101,000 in the extended social network - ie, you, your friends, and their friends. The chances that you're in some of these photos and don't even know it are very high - thus the instant popularity of Friend Finder," he tells ISRAEL21c.
Hirsch decided to work initially with Facebook because it was a good place to get started, but the technology goes far beyond identifying friends you don't know.
"Until now, nearly all photo recognition applications and technologies have been developed for the security industry - for example, a camera would compare the face of a visitor to a secure facility against a database. But photos taken by consumers and uploaded to internet sites are much more problematic," says Hirsch. "The photos are usually very poor quality, with shadows, red-eye, and all the other disadvantages."
Face recognition 'in the wild'
Face recognition technology for security purposes was designed to work in optimum lighting and atmospheric conditions - after all, the entrances to secure facilities are well lit, and the visitor has to look straight ahead, as per the security guard's instructions.
That technology, says Hirsch, worked very poorly with photos "in the wild," and the company realized it would have to start from scratch.
"Without getting too technical, our technology looks at information that is already known - photos in your Facebook account, for example - and compares them with elements of other photos with unknown elements. Our algorithms compare the photos, and Friend Finder makes an educated guess on the identity of a person. The user is then asked to confirm, and a tag is attached to the identified person, with that photo now added to the recognition database," he says. [MORE]