Help required

[TimberWolf]

Well done for not actually reading the link. FaceBook and Picassa both recognize and identify people.

Under extremely favourable and unrealistic conditions though. That is, a nice big face, square on, perhaps set against a plain background, good lighting etc, rather than a moving face in crowd seen from oblique angles, differing hairstyles, lighting, facial expressions and accoutrements. Just recognising people (without individual identification) would be hard enough under real conditions. Too hard probably.

[DaveB]

If you want to nip it in the bud ask her what it is supposed to do, google for it and find the dozen or so apps that already do that then ask her why hers will be better.

On the offchance that she's cmoe up with a genuinely new idea, your stuffed. get the wallet out and get coding

[Churchill]

Under extremely favourable and unrealistic conditions though. That is, a nice big face, square on, perhaps set against a plain background, good lighting etc, rather than a moving face in crowd seen from oblique angles, differing hairstyles, lighting, facial expressions and accoutrements. Just recognising people (without individual identification) would be hard enough under real conditions. Too hard probably.

Nope. The technology is out there.

I have a couple of cameras from these people on my desk as we speak.

[d000hg]

Under extremely favourable and unrealistic conditions though. That is, a nice big face, square on, perhaps set against a plain background, good lighting etc, rather than a moving face in crowd seen from oblique angles, differing hairstyles, lighting, facial expressions and accoutrements. Just recognising people (without individual identification) would be hard enough under real conditions. Too hard probably.

I would have said the same but it seems this has taken a massive leap forwards in just the last couple of years.

For an iPhone app, it doesn't have to be CIA-quality... just something that works OK and is fun.

[scooterscot]

I just find that the hype machine that goes with them grates a bit.

You know I've never advocated people should use a one computer over another or given any hype. It's a reputation I've been awarded for uttering the M word one single time many moons ago.

Window's users the world over, it's like mass Stockholm syndrome or something.

[TimberWolf]

I would have said the same but it seems this has taken a massive leap forwards in just the last couple of years.

For an iPhone app, it doesn't have to be CIA-quality... just something that works OK and is fun.

I think a lot of is hype and sales patter, though as you say things have likely improved over the years, and in specific or ideal settings result may be acceptable, e.g. for authentication rather than identification. Don't smile for your passport photos! I once wrote a basic image recognition system for furniture that worked well, but doing it for people and on a smartphone isn't something I'm likely to entertain, absent thermal images.

[doodab]

You know I've never advocated people should use a one computer over another or given any hype. It's a reputation I've been awarded for uttering the M word one single time many moons ago.

Window's users the world over, it's like mass Stockholm syndrome or something.

I wasn't actually meaning you. Although you do post "Get a Mac" about 10 times a week. I was referring to the entire Apple marketing machine. Like I said, they are good consumer electronics, but they are a good design & marketing company, not an innovative R&D company.

People who think Steve Jobs is some sort of genius single handedly reinventing computing for the 21t century are on a par with people who think Susan Boyle is the be all and end all of female vocals, IMO. Even an otherwise intelligent guy like NF has been taken in and seems to believe that the mac is some sort of magic from the future when in actual fact if it wasn't for the success of the "20th century" PC there wouldn't be any macs because without the cheap commodity hardware that PCs bought about they would have gone the way of DEC, SGI and half a dozen other "workstation" based on less popular architectures.

The future of computing isn't going to be sitting in front of a screen using a touchpad, it's going to be wearable, ubiquitous, reconfigurable. Apple might sell it well when it comes along but they sure as **** aren't inventing it and they never have, whatever they manage to persuade the beeb to report. It's that level of bulltulip with little or no credit going to a lot of true pioneers that annoys me, not the people who have fallen for it.

But like I said, they are good comsumer machines. If someone random asks me what laptop they should buy I tell them to get a mac more often than not

[doodab]

I think a lot of is hype and sales patter, though as you say things have likely improved over the years, and in specific or ideal settings result may be acceptable, e.g. for authentication rather than identification. Don't smile for your passport photos! I once wrote a basic image recognition system for furniture that worked well, but doing it for people and on a smartphone isn't something I'm likely to entertain, absent thermal images.

It'll be a lot easier when you just have to suggest the closest potential match from a pool of maybe 100 people.

[TimberWolf]

It'll be a lot easier when you just have to suggest the closest potential match from a pool of maybe 100 people.

Yeah, something like that would be easy, especially as the faces are presented nicely.

I had a quick look at what the state of the art was as written in peer reviewed scholarly articles, rather than sales patter or in unrealistic settings, but the first best one I found was from 2002:

http://citeseerx.ist.psu.edu/viewdoc...=rep1&type=pdf

The summary near the bottom gives a detection rate of about 90% using various methods and I'm not sure that recognising people (i.e their shapes) would be any easier than detecting faces, perhaps much harder (absent thermal cameras).

[realityhack]

Use this: mokk.me - Mobile web app mocks

Simples.