Slashdot Mirror
UK Police's Porn-Spotting AI Keeps Mistaking Desert Pics for Nudes (gizmodo.com)
An anonymous reader quotes Gizmodo:
London's Metropolitan Police believes that its artificial intelligence software will be up to the task of detecting images of child abuse in the next "two to three years." But, in its current state, the system can't tell the difference between a photo of a desert and a photo of a naked body... "Sometimes it comes up with a desert and it thinks its an indecent image or pornography," Mark Stokes, the department's head of digital and electronics forensics, recently told The Telegraph. "For some reason, lots of people have screen-savers of deserts and it picks it up thinking it is skin colour."
The article concludes that the London police software "has yet to prove that it can successfully differentiate the human body from arid landscapes."
The article concludes that the London police software "has yet to prove that it can successfully differentiate the human body from arid landscapes."
One hump, or two?
I've fallen off your lawn, and I can't get up.
""For some reason, lots of people have screen-savers of desert"
Are they really openly admitting to RAT-surveilling everyone?! What am I missing?
WTF are police doing scanning the images used for screen savers? Those aren't, generally, on a public server.
It the UK cops have a trojan on UK computers, I think the rest of the world should get to work burying their servers in noise. Antivirus definitions need updating ASAP.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
African, Asian, European desert lands all uncovered and untouched by man. Clean shaven of all foliage. Aching for a BBC (big black camera) to shoot all over their majestic vistas. They want long lenses and offer roads for deep penetrations into their wildest areas. Only $9.95/mo gets you all access to hours of video and thousands of pictures. Round, brown mounds of dirt, golden, flowing sands, fiery red sunsets are all waiting for you!
A picture of a nude person in a desert must render this thing catatonic ;)
Zo thinks about as well as Helen Keller did.
The big question is why the hell do they bother to detect nudes?
I would even ask why the hell do they want to spot pornography?
Both are very common human expressions and they should not be a reason for police investigation.
"The likes of Facebook and WhatsApp are free to those whose privacy is of zero value."
There is legit usage cases for this software. I, for one, would not trust people who jack off to images of deserts. They should be put on some police list of some sort, so that I know to avoid them.
Violence is the last refuge of the incompetent. Polar Scope Align for iOS
I can understand the problems they're having....
If you gave me a choice between a printer and a giraffe with explosive diarrhoea, i'll get my ladder and my raincoat
Racists!
I am armed because I am free. I am free because I am armed.
DUNE is a misspelling of NUDE.
Perhaps this says something about the obesity epidemic...
I mean, correct me if I am wrong but doesn't the only way to see what images people have on their desktops mean that the Metropolitan Police are poking around and doing some sort of unauthorized remote access into peoples' machines?
How else would they become aware of the types of screensavers people are using? Not sure but I find this more than slightly creepy.
Is this more of that famous Nanny State® at work again? "Think Of The Children" gone wrong...
is delicious! :)
Once you've fed "Paradise (1982)" into the AI, all arid scenes will trigger that response. It doesn't matter the age. Same as humans.
No, they're not trying to get the computer to do all the police work, they're just trying to use it to narrow it down to nude images so that the amount of work they have to do is much smaller. This is for searching people who are already suspects.
I came across an extremely simple skin detection algorithm some time ago and made a GLSL implementation. Check here for a demo and explanation (click about). http://labs.thygate.com/amazin... It is actually quite robust if the lighting conditions of the scene don't change too much.
Generally speaking they're not going to be trying to distinguish a 17 year old from an 18 year old. They're going to be trying to distinguish a 10 year old from an 18 year old. And there's definitely some obvious physiological differences there (in most instances.)
What are they thinking?
banana split
first, Christians. There's a concept I call "Aggregate Sin" because I can't think of a better name for it. Christians (the ones who take the Bible literally) believe that if there's too much sin in the world God will punish us. e.g. God punishes the Faithful for the sins of the heretic. Therefore sin, any sin, is an existent threat to them. If you extrapolate from the tales of Sodom & Gomorrah & Noah's Ark this makes sense. You see this most commonly when a natural disaster strikes and a bunch of them come out of the wood work to claim it's the Lord punishing us for sinning. The important thing here is to understand that the Lord isn't very picky about who he punishes. This incentivizes Christians to control non Christian behavior; up to and including converting them. It's a big part of why Christianity spreads the way it does.
As for the other reason, well the entire system is designed to keep the working class' attention somewhere else while the Aristocracy pockets all the money. It's a classic technique (along with Balkanization, usually along racial or caste divides but sometimes along ideological, e.g. wedge issues). Basically, if 1% of the populace is going to take 50-90% of the wealth for themselves they have to create all sorts of crazy systems to make that work.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
Allow me to further elaborate as to WHY i found this algorithm so interesting, because it's so ridiculously simple. It considers a pixel is skin if 50 R - G 250, where R and G are the red and green channels from a 8-bit RGB pixel representation. Sure it's not meant for the prime-time, it's just a very interesting curiosity, and easy to perform in parallel.
darn html got filtered of < symbols.
Damb there in there somewhere... all 70 of em.
[($)]
It has gone beyond dreaming and is seeing mirages!
...in you are into that.
You just think it's easy because you're human and take for granted how very subtle the visual cues are.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Maybe it's dyslexic, and can't tell the difference between small nudes and small dunes.
There's no time like the present. Well, the past used to be.
https://www.sadanduseless.com/...
hot sandy mounds
That's kind of my point. Its not easy, even for a human, to distinguish a 17 year old from an 18 year old based on pure visual clues. But other than extreme cases of precocious puberty its usually pretty easy for a human to tell a 10 year old from an 18 year old. Meaning that at least in principle, the latter should be doable by a smart enough AI (which is not saying such an AI is necessarily easy to build of course. Just that it should be possible based on how easy it is for us humans to do the job.)