The Era of Automatic Facial Recognition and Surveillance Is Here • Hacker News Board

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#111

Post by terryf » Tue, Oct 06, 2015, 6:16 AM UTC

Houshalter wrote: post #78

macNchz wrote: post #51

Earlier quoted context omitted.

You've spent years saving up to finally buy a home. You're having a baby in a few months, and you really need the space. You have the down payment ready, your credit is great, you've ticked all the boxes on the application, but you're turned down for the loan. Why? Maybe you walk home through a bad neighborhood. Maybe you sometimes grab a beer at a bit of a divey bar near your office. Maybe you've visited an old frie…

On the other hand you are just as likely to get a loan you otherwise wouldn't have because you haven't been seen with questionable folks. In any case that kind of policy is already illegal. And it's not even technically feasible at the moment. Even if it was, it's very unlikely there is a strong correlation between being seen with some random person in a photo, and your likelihood of paying a loan. There is plenty of…

The banks are doing things, perhaps not exactly what was described, but your facebook history, linkedin, other social media accounts are taken into account when applying for a loan. Even when you browse the bank's web page, where you hover the mouse, what pages you look at and variour other metrics. One example of a company providing such service is http://www.bigdatascoring.com/

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#112

Post by Silhouette » Tue, Oct 06, 2015, 6:19 AM UTC

cookiecaper wrote: post #37

hackuser wrote: post #7

Earlier quoted context omitted.

> surveillance is here to stay and that making it illegal will only serve to hide the microphones. Certainly it's here to stay, but so is murder, theft, fraud, and many other things. We want to maximize the leverage and tools people have to fight it.

Unfortunately you can't put the genie back in the bottle. I don't think there's a real way to fight this in the age of networks and cameras in every pocket. There are many good uses for the ability to instantly recognize and identify anybody from a crowd, as well as bad ones. Probably the easiest way to frustrate these things would be the simplest; only go out in public with substantial facial features obscured or al…

I don't think there's a real way to fight this in the age of networks and cameras in every pocket.

Sure there is.

For one thing, a similar level of technology to the capture and processing applications could also be used to automatically detect and obscure faces where storing them isn't necessary, probably with imperfect but highly reliable success. Consider how Google's Street View wound up hiding car licence plates for a lot of places after the privacy complaints, for example.

For another thing, the problem with a lot of modern privacy invasions is asymmetry: large, powerful organisations like big businesses or governments can do a lot more to you with lots of information than you can do to them in return.

This could be solved for businesses by imposing rules and meaningful penalties for violating them. Facebook isn't going to be processing anyone's face for dubious reasons if it's subject to compulsory scrutiny on account of processing a high volume of personal data and the penalties for repeated breaking of the rules escalate as significant percentages of annual revenue. For deliberate violations, you could remove the corporate shield from the responsible executives as well, as with other criminal conduct.

Governments is harder, because right now the governments of the West seem to be more hostile to their own people than to foreign surveillance. Ideally, our governments would actually be on our side, and the vast resources they seem to be devoting to invading their own people's privacy would instead be devoted to protecting it. Imagine how much safer we would all be from things like fraud, identity theft, commercial espionage, and worse, if instead of trying to exploit vulnerabilities and limit secure communications, organisations like the NSA and GCHQ helped to close vulnerabilities and promote safe and secure use of technology by everyone. Like everyone else, I don't know exactly what the real threat from terrorism is, but I'm pretty sure the potential savings to our economies alone from safer technology would run to many billions.

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#113

Post by terryf » Tue, Oct 06, 2015, 6:20 AM UTC

bostonpete wrote: post #68

analyst74 wrote: post #60

Earlier quoted context omitted.

But doesn't this cut both ways? If you don't hang around questionable people, you'd qualify for better rate for loans. Or if your credit score is not very high, but the fact you don't hang around questionable people brings your credit file up a notch and make you qualify.

Yeah, in my mind, this completely negates the argument. If I don't engage in a given statistically risky behavior today, I'm paying more for insurance because some people do and the insurance company doesn't know who should be paying for that risky behavior so we all pay a little more. To me, the more compelling angle to attack surveillance has to appeal to people's desire to keep some portion of their lives from bei…

But the whole point of insurance is to collectively spread the risk, so that when things do go bad, you are not on your own. By definition, people in aggregate pay more for insurance than what the insurance pays them back (otherwise the company would go bankrupt!).

Insurance is paying for peace of mind. Most people are paying more premiums than insurance is paying them back.

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#114

Post by yorak » Tue, Oct 06, 2015, 6:21 AM UTC

meric wrote: post #88

duderific wrote: post #77

Earlier quoted context omitted.

Ha, when watching those shows, I am always the nerd telling my wife how it's impossible to zoom in and enhance, because you can't create more information than existed when the photo was taken. I'm not sure that she gets it.

If you take a blurry photo of a license plate - it can be possible to enhance it (e.g. using sharpen) to reveal the license plate number - just because your naked eyes can't extract that information from a blurry photo doesn't mean the blurry photo doesn't already have enough information to encode the number precisely.

There is also superresolution methods that in practice make the "enhance"-feature possible. The information from the same objects in successive frames, eg. from a security camera feed, are composited into a single frame of higher resolution than the originals. This can even be done for a single image (see. Glasner et al. "Super-resolution from a single image." Computer Vision, 2009 ICCV 2009. IEEE.).

If machine learning is brought into play it gets even more interesting. See eg. waifu2x, the anime video upscaler based on Deep Convolutional Neural Networks. Basically it is a neural net, that has been shown A LOT of anime, and it can therefore creatively(?) fill in the created pixels when upscaling.

I have no idea how advanced this kind of ML boosted "enhance" technology can get in the future, and it might even get a little scary: imagine an observer with the experience of million lives spent looking blurry images in poor lighting. It might get surprisingly good at that...

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#115

Post by Pxtl » Tue, Oct 06, 2015, 6:25 AM UTC

macNchz wrote: post #51

hackuser wrote: post #6

A HN contest: Provide an elevator pitch, to completely non-technical end-users, that will make them aware of this issue and persuade them that it is important to them. Vote up the best ones. [EDIT: The goal is to persuade them about its dangers, not sell them something based on it] That elevator pitch is key. I have extensive knowledge about this issue but have a hard time coming up with it; I need to be prepared. Ch…

You've spent years saving up to finally buy a home. You're having a baby in a few months, and you really need the space. You have the down payment ready, your credit is great, you've ticked all the boxes on the application, but you're turned down for the loan. Why? Maybe you walk home through a bad neighborhood. Maybe you sometimes grab a beer at a bit of a divey bar near your office. Maybe you've visited an old frie…

This isn't a good example because the technology already exists to deliver this service to lenders - read up on Canadian Tire's JP Martin. He did the first experiments with rewards-credit-cards and data mining, and found he could pretty-well correlate certain behaviors with missed credit-card payments. Apparently there's a dive bar in Montreal that indicates you're a screw-up if you drink there.

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#116

Post by jaxb » Tue, Oct 06, 2015, 7:04 AM UTC

devindotcom wrote: post #26

There was a comic recently that had a main character who had little emitters constantly blasting his face with IR radiation, so it would just turn up as an overexposed blob on surveillance cameras. Of course all they need is an IR filter to combat this, but the idea is fun.

an episode of Almost Human had 'anti-camera spray' that did this, as well. maybe a comic was based on that?

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#117

Post by nitrogen » Tue, Oct 06, 2015, 7:06 AM UTC

rm999 wrote: post #95

jpatokal wrote: post #65

So are we actually there yet? I was under the impression that, while face recognition has evolved to the point that face matching against a limited group of people (say, your Facebook friends) is feasible, we're still a long ways off from being able to put a face-scanner in a crowd in NYC and have it spot the 1 suspect among 10 million without a totally ridiculous rate of false positives.

This should be a top comment - this kind of tracking is not possible at scale and probably won't be for a long time. Identifying a person in a group of 100s/1000s has been a solved problem for decades. Consistently identifying a single person from a database of 100Ms/billions of people who are constantly gaining weight, growing facial hair, etc is impossible with today's technology; false positives will become an exp…

There are already devices that can rapidly scan irises of a crowd. It's also possible, as the sibling comment suggested, to correlate multiple datasets (wifi probes, unique phone CDMA/GSM/LTE IDs, location, behavior, physical limitations like maximum velocity, etc.).

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#118

Post by finance-geek » Tue, Oct 06, 2015, 7:17 AM UTC

rm999 wrote: post #95

jpatokal wrote: post #65

So are we actually there yet? I was under the impression that, while face recognition has evolved to the point that face matching against a limited group of people (say, your Facebook friends) is feasible, we're still a long ways off from being able to put a face-scanner in a crowd in NYC and have it spot the 1 suspect among 10 million without a totally ridiculous rate of false positives.

This should be a top comment - this kind of tracking is not possible at scale and probably won't be for a long time. Identifying a person in a group of 100s/1000s has been a solved problem for decades. Consistently identifying a single person from a database of 100Ms/billions of people who are constantly gaining weight, growing facial hair, etc is impossible with today's technology; false positives will become an exp…

"people who are constantly gaining weight, growing facial hair, etc is impossible with today's technology;"

I think we can assume many agencies have access to ALL Facebook data, Photostream data, etc. With that it is very much possible, dare one say, easy?

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#119

Post by finance-geek » Tue, Oct 06, 2015, 7:19 AM UTC

bigiain wrote: post #96

jpatokal wrote: post #65

So are we actually there yet? I was under the impression that, while face recognition has evolved to the point that face matching against a limited group of people (say, your Facebook friends) is feasible, we're still a long ways off from being able to put a face-scanner in a crowd in NYC and have it spot the 1 suspect among 10 million without a totally ridiculous rate of false positives.

You say that as though the people deploying things like this[1] care much about false positives... [1] No fly lists, for example.

I can verify they do not care about false positives. My high-flying career at Accenture ended abruptly in 2004 when I ended up on the no-fly list. Didn't get fixed until 2009, and fixed is sort of gray.

Re: The Era of Automatic Facial Recognition and Surveillance Is Here

#120

Post by finance-geek » Tue, Oct 06, 2015, 7:29 AM UTC

AnimalMuppet wrote: post #44

gknoy wrote: post #42

Earlier quoted context omitted.

I think it's unwise to underestimate your fellow humans' propensity to "be a dick", as you put it, even if they don't gain anything more than the satisfaction of treating others poorly. Facial recognition + a database allows POWERFUL and fine grained discrimination! Imagine being able to know ${Things} about people -- via analytics / The Internet -- whenever any prospective customer comes into your store. You could r…

> I think it's unwise to underestimate your fellow humans' propensity to "be a dick"... That tendency exists quite strongly in humans, yes. Unfortunate, but true. On the other hand, businesses also have the tendency to want to make money. If I've got a business, and I make it hard for group X to buy stuff from me, they're probably going to try to find someone else who wants more strongly to make money and/or less str…

"And there may be businesses that are okay with that. But it ought to be somewhat of a self-defeating behavior."

This type of libertarian self-solution rarely works. Things do not self-defeat because the affected groups are small and grow slowly. Consider the restricted fly list -- part of it was was government lead but part of it was implementation lead -- companies decided their own filters. Some were horrible -- you selected "Muslim Meal" option and the next thing you know you cant get a boarding pass. Problem is, you dont always have options. If you got restricted form American Airlines, as I did, suddenly you found yourself with half the flight options. If you got flagged again by United, now you were in real trouble...are you going to fly from NY to Charlotte via Detroit?