Hacker News new | past | comments | ask | show | jobs | submit login

Very interesting article. Too bad google glass didn't catch up, or we could set up reasonably inconspicuous face recognition to help these people. Otherwise, I guess they could have a hidden camera somewhere on them and a (wireless) earbud that tells them who's in front of them.



That sounds like a much better solution than google glass, and one that we could even build in a weekend.


I think it would be harder than that. Bear in mind that this will use images of moving targets from strange angles in poor lighting. Is there any system that works that well, even out of something like Facebook or Google? Plus this has to be portable and battery-operated.

I know nothing about computer vision.


the real issue is where to mine the names and pictures from. Facebook? LinkedIn? The tighter the pool of people the best the face recognition will be. Asking people to enter everyone in by hand is probably unreasonable, though I'm not sure


I think just the system recognizing faces that it's seen before would be good enough. Entering by hand isn't too bad.

When it detects a new face, add it to the "seen faces" list, and let the user go back though that (sorted most recent at the top) whenever they like, to add names and maybe a short note like "met at party in LA". So you have to remember their name yourself but only momentarily until you can enter it into the local DB.

Whenever that face is detected again, it shows the info.

That way you can skip Internet connectivity on the device entirely.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: