As Wired writes, it is a conceptual piece. I see the vision technology comes from Polar Rose. I've worked with them for a couple of years as their lead architect. Until they funding dried up. I think they are running on fumes now.
Vision technology is really tough to sell. Finding faces is simple, everybody is doing that these days, but actually recognizing a face is a whole different game. It is really hard and you need a lot of cpu power and storage to match someone in near realtime. At Polar Rose we got good results with stock photos. But throwing 'internet quality' photos at it resulted in just terrible accuracy.
What they do in the demo cannot work just on the phone. It will need a support infrastructure that receives the mugshot, processes it and then matches it against other faces in the database. Nobody really gets this part right yet.
I used to develop some facebook apps. I think in this aspect you could process your network (and maybe their networks?) pull all the photos of those people and look for their face. Facebook doesn't share the "tagging" data afaik, but you could build a limited library of who you're going to be taking photos of. So even if the recognizing isn't very accurate it wouldn't have to be precise.
I was once approached to help build something like that for a prison. Essentially, it'd match the prisoner to their mugshot to ensure the right person was let out. I turned the position down, as when I asked about how they would store the eigen-vectors for that many people and index them I was met with an awkward silence.
I do like the idea of processing Facebook's data, but that's another plan entirely.
Looks like Augmented Reality = your inner self projected online.
Or, your many inner selves living an online, second life.
I was thinking many times What Is The Best Personal Business Model? I've realized alone I can't compete with one or more people grouped in a team having the same goals.
Then I've learned a cornerstone of competitiveness is having (more and more) people working for you, or at least not working for your competitors.
I've also learned managing more and more people is not fun so I was starting to build a team of online applications working for me which is easier and more fun to manage.
But the future is to project your inner selves into the digital reality, make them living and working on their own, and meet them frequently to re-live their experiences. (and cash their revenues of course :-)
It's something like cloning yourself not biologically but digitally. Perfect.
Vision technology is really tough to sell. Finding faces is simple, everybody is doing that these days, but actually recognizing a face is a whole different game. It is really hard and you need a lot of cpu power and storage to match someone in near realtime. At Polar Rose we got good results with stock photos. But throwing 'internet quality' photos at it resulted in just terrible accuracy.
What they do in the demo cannot work just on the phone. It will need a support infrastructure that receives the mugshot, processes it and then matches it against other faces in the database. Nobody really gets this part right yet.