Tampa Ditches Face Recognition After It Catches Exactly No One In Two Years
from the sounds-useful dept
There was a huge uproar after police in Tampa started using facial recognition software to try to catch wanted criminals. More than two years after the system debuted, police are tearing it down saying that it didn’t help them catch a single person. It turns out that, the face recognition system pretty much saw everyone as being regular people (well, indirectly it did catch one person – but it was the wrong guy, and it was because a demo of the system caught the eye of a woman who insisted, falsely, that the guy seen in the demo was her ex-husband). While plenty of people have said for years that face recognition systems for catching people in a general crowd are useless, now there’s a bit more evidence to support that assertion. A police spokesperson still tried to put a positive spin on the test saying that maybe knowledge of the system kept criminals away from Tampa. Yeah, I’m sure that was the first thing on the minds of fugitives on the run: “gotta avoid Tampa, they have a facial recognition system on the streets…”
Comments on “Tampa Ditches Face Recognition After It Catches Exactly No One In Two Years”
What about 2nd-generation facial recognition?
The current generation tried to match 2-D pictures of faces by use of “eigenfaces”, hence its inaccuracy.
Next-generation scanners can do invisible 3-D scans by infrared lasers, which will be much more accurate.
Re: What about 2nd-generation facial recognition?
Agree. More processing power, storage, and methods of gathering data, and this will be possible. Needs some work in the lab before being ready for prime time, but it will be here during my lifetime.
Looks like we gotta go back to posting pictures of our most wanted at the post office – No replacement yet for real people (see other posts about robots replacing humans).
Sorta OT – did anybody read those articles about actors who played crooks on “America’s Most Wanted” actually being detained after being spotted by people in crowds? Humans aren’t that accurate on face recognition, or if they are, don’t necessarially have the right data.
Re: Re: What about 2nd-generation facial recognition?
Facial recognition is easily fooled.
Some days I have a goatee,
others I have a mustache.
Once in awhile, I’m clean shaven.
Sometimes I wear a baseball cap or a ‘do rag.
Today my eyes happen to be brown and I am wearing glasses.
Tomorrow I might wear my green contacts.
Last week my hair was a lovely shocking yellow.
Two nights ago I shaved my head.
A week ago I dyed my skin a beautiful shade of tan.
All because I enjoy changing the way I look.
If my friends don’t always recognize me, why would anyone think a 2nd generation machine will ?
Does anyone SERIOUSLY think one of these machines would positively identify Micheal Jackson, Then & NOW ???
The American Spirit will always find a way to spoof the system.
And no, I’m not a criminal, I just don’t particularly like the idea of machines ( of far less worth than humans ) making critical decisions as to whether or not I’m worthy of whatever they are enslaved to proportedly protect.
Re: Re: Re: What about 2nd-generation facial recognition?
You know facial recognition systems don’t rely on colour at all. They utilise distinctive features of the face – including the upper outlines of the eye sockets, the areas surrounding the cheekbones, the sides of the mouth, and the location of the nose and eyes – to perform verification and identification. So most of your points don’t count one bit. However, the mustache may come in the way if it’s really bushy and possibly a lot of hair on your face may as well.
But I do see your point – like say someone with light sensitive eyes just had to wear sunnies all the time. And that isn’t too uncommon. They are also really really inaccurate with about 80% accuracy with a false positive rate of 1%.
Do your research next time.
Re: What about 2nd-generation facial recognition?
Next-generation scanners can do invisible 3-D scans by infrared lasers, which will be much more accurate.
That’s all fine and good, but where will the police get 3-D pictures of criminals to compare the live scans with? Your only as good as the data you’ve saved previously.
What about 2nd-generation facial recognition?
I know of at least 2 councils in the UK that have tried and failed with facial recognition technology. We’re the most watched nation on earth so we know a thing or two about what’s possible. The reason being that it only ever works if the subject is compliant when the original scan is made, and again when they are captured on CCTV. Since this never happens in real liife – and bad people have a habit of wearing hoodies and shades – the system has never worked in general street scene applications. Let me repeat – facial recognition ONLY works when the subject is compliant. That normally means sitting them down initially for the scan and later asking them to remove headwear on passing through a camera gate no more than a few at a time. When vendors come back at you with 3D scanning they’re missing the point. There’s no point employing Leonardo Da Vinci to copy a fake painting….right.
What’s also interesting is that the only reason that cameras are accepted here in the UK is that. like guns, we don’t trust them in the hands of our police forces. They are all operated by local authority employed civilians under very strict data protection guidelines. Incidents are then reported onwards to plod – who STILL says he’s too busy.
IMHO – Behavioral recognition from companies like Ipsotek has a lot more potential.