Facebook's Face Identification Project Is Accurate 97.25% of the Time 149
kc123 tips news that 'DeepFace,' the software research project created by Facebook engineers to identify people in pictures, is now accurate 97.25% of the time. In other words, it's almost as good at recognizing faces as humans, who are able to determine whether two photos show the same person 97.53% of the time. The article says DeepFace reaches that level of accuracy "regardless of variations in lighting or whether the person in the picture is directly facing the camera." It continues,
"DeepFace processes images of faces in two steps. First it corrects the angle of a face so that the person in the picture faces forward, using a 3-D model of an 'average' forward-looking face. Then the deep learning comes in as a simulated neural network works out a numerical description of the reoriented face. If DeepFace comes up with similar enough descriptions from two different images, it decides they must show the same face. ... The deep-learning part of DeepFace consists of nine layers of simple simulated neurons, with more than 120 million connections between them. To train that network, Facebook’s researchers tapped a tiny slice of data from their company’s hoard of user images—four million photos of faces belonging to almost 4,000 people."
Re:Say goodbye (Score:5, Interesting)
Muslims are right: Burqa's are the way to go...
Re:Say goodbye (Score:3, Interesting)
No, start protecting your privacy.
You do not have to allow then to do it.
http://cvdazzle.com/ [cvdazzle.com]
And, self promotion (baby steps, but working):
https://play.google.com/store/... [google.com]
Re:Privacy nutjobs take note (Score:4, Interesting)
You are blind to the fact that this is not a matter of going online to facebook. This is the reverse. This is the point where facebook starts coming to you. In real life. In the street, at the airport, in the store, at the dentist. And it'll know you, not necessarily because you've told it, but because all your acquaintances have.
Facebook will pass that information on to the airport/store/hospital because they'll pay to know who you are before you even approach the counter. "So they can better serve you."
Math (Score:4, Interesting)
Training the data (Score:5, Interesting)
I logged into Facebook for the first time in about 6 months, and it required me to authenticate myself by answering a series of questions about who was in each picture. It would display 3 pictures, each showing a square around a particular person, and it would ask who the person is. It was multiple choice.
I wonder if this is how they confirm that the data is correct, to eliminate intentional errors. You can ask a person who doesn't own the picture and didn't tag it to confirm the person in there. By masking it as an authorization request you convince people who otherwise would not be involved in tagging to participate.
Why this DOESN'T work (Score:4, Interesting)
Recently, FB decided that it needed to verify that I was really me when logging in. To do this, it presented me with a bunch of photos from my "friends" that had been tagged and insisted that I choose a name of someone in the photo. If I got enough of them wrong, it would "lock" my account. (Not quite "lock" but I had to try it again). Not only did it pull up obscure photos from "friends" I rarely interact with so I had little chance of knowing who was in the photo. But get this: It pulled up photos of people facing away from the camera and expected me to know who the person was from behind. Da fuq, FB? Seriously?!?
Re:Say goodbye (Score:4, Interesting)
When this happens, think of the convenience! All you'll need to do is look at the nearest camera and give a thumbs up, and Facebook will automatically mark that you Liked [whatever you're standing near].
Two people could become friends by finding the nearest Big Brother station and doing a thumbs up together.
One of (many) problems will be how they contextualize all that data. You know, this started as a joke, but seriously, if Facebook had a feed of this kind of data, it would be interesting to see the hypothetical profile they build based on what they would see an individual near vs. what they claim to like on their public page.