"Before the advent of these new technologies, time and effort created effective barriers to surveillance abuse. But those barriers are now being removed. They must be rebuilt in the law."
Time and effort maintained privacy in the past, by default. However, as technology improves, privacy is no longer the default, so we need to request it, specifically.
I hear again and again that "no one cares about privacy", but I'm starting to think this is a relic of the past. Perhaps 10 years from now we will say "no one used to care about privacy, but that was before technology became sufficiently advanced."
Which the machine won't recognise as a person (and will discard). It would take a human to track you but that's old school. Of course, that's only until gait recognition takes over from face recognition.
haha ... I totally understand that GP didn't mean it this way. But there is something really funny about that statement, as if walking around with a bedazzled face won't make it easy to just say, "the person with the bedazzled face" to describe a very specific person :P
Given the rate of false positives when "cold hits" are used with DNA databases, facial recognition (which is presumably higher entropy than DNA) has the potential to send a lot of innocent people to jail.
I also have some concerns of inadvertent "software racism". See for example, this video: https://www.youtube.com/watch?v=t4DT3tQqgRM. Obviously facial recognition isn't inherently incapable of recognizing people of different races, but if it's designed and calibrated on subjects of one race, it won't necessarily work well for subjects of different races. The last thing we want is a facial recognition database that thinks e.g. all Arabs look alike.
And of course, there are a ton of ways that it can be thwarted easily and inconspicuously, even if it works very well. Are all surveillance cameras going to be fitted with IR filters to ensure that a bright IR LED necklace doesn't blind them? And how hard is it for a criminal to hire a makeup artist?
If VR headsets like Oculus Rift take off, this problem can be mitigated. People will walk around with headsets to cover their entire face, defeating the cameras.
Surveillance is going to happen, if you ban government surveillance, the public will get this tech and drones and be doing it anyway. So the robust way to fight it is to plan your life knowing that you can always be monitored and someone will always know where you are, even at home.
This is a great opportunity for startups. Webapps that allow you to plan your security, kind of mini-fiefdoms with your family, friends, neighbors, associations that plan against attacks from rivals. Electronically activated weapon systems that shoot when a threat is detected. The Mad Max era is coming, and hackers will get many opportunities to earn a fortune.
Because it wasn't clear to me at first, I'll just point out that the pilot for this thing is using a VR headset similar to Oculus Rift, and the video is sent back to him over wifi. Supposedly the range is on the order of a mile or so.
I didn't even realize this was a feasible thing already! Looking into it, you can get these things for only $1500 or so. And so suddenly I've just realized, 10 or 20 years from now, it's not just the police who will be flying these things around. Why not buy one to use for your own neighborhood watch, right?
Hell, I'd love the chance to fly one of these around just for fun.
And so suddenly I've just realized, 10 or 20 years from now, it's not just the police who will be flying these things around.
On the contrary. In a year or two, it might not be just the police. In 10 or 20 years, I expect they will be regulated, a licence will be required to operate one, and they will be required to transmit identifying information at all times. Apply any argument you like that relates to driving on public roads, any argument you like that relates to handling potentially sensitive personal information, or any argument you like to do with navigating crowded airspace.
Perhaps, but the cameras on the VR headsets will provide even more data for the government for those who don't have an anonymous face; especially if the data is sent to a central server for processing (ala Google Glass).
Plus, unless you build the hardware entirely yourself, the headset won't save you from other tracking technologies. Presumably the headset will have a wireless chip whose MAC can be tracked, or, worse, it'll have an RFID chip specially-built for government tracking.
People will walk around with headsets to cover their entire face, defeating the cameras.
Let's face it, you're won't going to walk around with a VR headset on your face unless you're a) wireless connected to a server, or b) enjoy taking a heavy backpack or cart of gear everywhere.
1. I know one thing. I am tired of my picture taken
whenever I walk into a store. In order to buy a bolt
at that mess of a store(the big orange); my picture is taken multiple
times. It pisses off customers like myself, and every
year for as long as I can remember, the majority of
theft is internal. The worst offenders are Management.
I think you will see a rise in people wearing desquises?
Or, maybe an Indian will eventually sue the offending
store on religious grounds? (steal the soul)
Yea, I'm rambling, but I sometimes feel like we lost all
privacy? And yes, this is the only site I comment on
anymore. I don't even trust this site. I bet they are
logging in all IP's, and saving them just in case the government comes a knocking?
I feel rather similarly about number plate scanning for cars. My local supermarket lost literally £10,000+ when they installed scanners on their car park a while back and we took our custom elsewhere as a direct result. I don't know how much they make in fines, but I expect my family alone cancels out everything they made for a while after introducing the cameras. (And yes, some of that chain's stores did get pulled up for trying to charge fines that weren't justified after installing similar cameras. It made the local paper, with resulting PR about as good as you would expect.)
My chief worry is overconfidence from hopelessly technologically ignorant bureaucrats combined with a slew of false negatives and false positives when applied to human faces on general surveillance feeds.
In contrast, driver's license, ID, and arrest photos are at least severely constrained input filters wherein a single individual is placed against relatively uncomplicated background pixels and the face is in roughly the same orientation.
I don't know about the cat detector, but Picasa seems able to identify people in photos pretty well, despite varying positions and lighting conditions.
It creeps me out.
Granted, they are working with the much smaller set of people I know rather than the whole populace. But technology only gets better.
Technology is bounded by mathematical constraints, though. The more you constrain the number of faces you are scanning and you are looking for, the easier the task is. But when you're trying to pick out any of perhaps thousands of persons of interest across an input set that is, say, in the millions, the more impossible it becomes.
If it helps, imagine that you've equipped a person with all the people they're looking for, and then set them up in front of the cameras. Even if you imagine a superhuman who somehow has time to examine all of the millions of inputs, even the human is going to have a huge number of false positives and negatives. The problem itself is fundamentally, mathematically hard. Any surveillance technique based on the idea that computer vision is better than human vision at this sort of thing is going to fail, hard.
And that's before the population starts taking active countermeasures against the face metrics. (As usual, despite the abundant evidence we live in a dynamic, reactive universe, most humans persist in functioning in a static model, where the Bad Guys will just passively sit and let their faces be scanned and never react to that.)
Just like there was a relatively cheap countermeasure to every pipe dream revision of the SDI, such countermeasures are already in place for facial recognition:
Those aren't even the interesting countermeasures. Good work with makeup can significantly change the structure of your face, completely reversibly, far more than is necessary to fool these systems running across the full set of human beings. You don't have to get the face recognizer to not recognize a face at all, you just have to make it a wrong face.
Identifying people's faces, especially caucasians, is relatively considered straightforward nowadays. Google spent a lot of time working on the problem so they could obscure faces in Google Street View. You start by looking for the eyes, which are dark spots surrounded by white. Then, you go from the eyes to the nose.
Identifying which person a face belongs to is a classification problem, which gets difficult the larger the set you're comparing to. Facebook has some tech that is interesting here, but it also only checks your friends and some level of false positives is acceptable. I imagine it that what algorithms they have aren't quite good enough to, say, identify your face out of an entire city's worth of driver's license photos. (Yet, which is what this article alludes to.)
> "While this sort of technology may have benefits for law enforcement (recall that the suspects in the Boston Marathon bombings were identified with help from camera footage), it also invites abuse."
1. Yes /without/ this technology, and
2. How would that have prevented the Boston tragedy if it were in use?
When they identified the first blurred images of the suspects' faces they put out the images hoping people could help identify the faces they were interested in. I assume they're saying that with better technology they'd be able to instantly identify those faces, instead of relying on putting them up on TV (where the suspects can see them and know they're on to them). So it wouldn't have prevented the attack, but it theoretically would have made the identification easier, and the arrests faster/less deadly I suppose.
Perhaps, but they were caught pretty fast already, to be honest. The insinuation is that somehow this technology would have prevented the attacks in the first place, which is highly improbable.
> "But as surveillance technology improves, the distinction between public spaces and private spaces becomes less meaningful."
Or more to the point: as the NSA mandates access to all private data, the distinction of recordings from public spaces and recordings from private spaces is made moot.
Yes, it exists, and I use it every day. In fact, I have a partnership with the DOD/NSA subcontractors who created it, and I have applied the technology to the automated creation of 3D Avatars given a single photo of a person's face. My version of the tech is available at www.3D-Avatar-Store.com. The system works as follows: a person's head is laser scanned, at the same time dozens of single photos are taken of that person from different angles, different lighting conditions, and different quality cameras. Then a neural net is trained to associate each photo with the laser scan data. After a few thousand trainings, anyone's single photo would generate a reasonable likeness of them in 3D. Today, after over a decade of training and 10's of thousands of scans added to the training database, we get remarkably good quality 3D reconstructions given a single photo of anyone, any age, any ethnicity. There are limitations, such as the face needs to be visible, and it needs to be within +/-30 degrees of facing the camera for the 3D reconstruction to be suitable for facial recognition. And that is the point of the original application of this technology: a facial recognition pre-processor. It corrects the face angle and removes any facial expression, creating a likeness of the face in a passport style photo perfect for facial recognition. But, of course, I'm not using it for that. I use it to auto-magically create 3D avatars for video game studios.
Oh, it's really, really fast. My current system is based upon two generations back of neural nets, and they reconstruct from one photo in 0.9 seconds. The latest generation does 144 reconstructions simultaneously given HD video feeds.
Very interesting technology. I used your website, uploaded three pictures -- it failed on the first, and worked somewhat on the others. I'm guessing it needs a more experienced user base? Assuming that model generation requires human supervision, do you think the neural network algorithm scales to the numbers mentioned in the article (trained on 12 million photos, and operating in 7 different states)?
BTW, all the best with your website. Hope you take more avatars "where go where no man has gone before!"
What's the state of the art in terms of extending such systems to handle faces that are partially obscured (e.g. wearing glasses)? References to further resources in the area accepted and appreciated! I'm going to start work on a face recognition system soon and I'm keen to learn more about their current limitations.
It's just an issue of training. I'd expect a similar setup as I describe could be used to associate a series of faces with and without different things attached to their face, and a neural net could be trained to generate, for example, a no eye glasses wearing image from one with eye glasses. We already do that with facial expressions. I think the issues to be improved upon are extending the recognition beyond the face - include ear characteristics to extend recognition beyond +/- 30 degrees, and although my lab tells me analyzing gait is a dead end, I feel there might be something there that can help. Also lighting neutralization is an area that needs some help, which includes illumination balancing without removing important details, shadow removal, and colored illumination white balancing.
It could certainly narrow the field. Combine a vector of "features" derived from the single image that with a bunch of other orthogonal "features" such as location and I could see it improving efficiency of manhunts significantly.
Maybe not in realtime, but they could store all the images and maybe have software presort faces by similarity. Then if they became interested in a specific person, they could input more training images and sift through the set of similar ones.
They might find hundreds of pictures of you with timestamps and locations.
(This is all speculation; I haven't written image recognition software myself.)
Why does it need to be a single photo? Mandate that when you get your passport/driving license theres a face scan. Wouldn't be amazing tech and could be introduced fairly easily.
I didn't realize that. Got my passport long ago, and for the driving license I'm sure they just took a picture. There was no scan (atleast nothing like the laser scan described by bsenftner above).
I do it with only a single ocular image (eye, eyebrow, and surrounding skin texture) rather than the full face. Using 1:1 matching environments are common for NIST challenges.
"Before the advent of these new technologies, time and effort created effective barriers to surveillance abuse. But those barriers are now being removed. They must be rebuilt in the law."
Time and effort maintained privacy in the past, by default. However, as technology improves, privacy is no longer the default, so we need to request it, specifically.
I hear again and again that "no one cares about privacy", but I'm starting to think this is a relic of the past. Perhaps 10 years from now we will say "no one used to care about privacy, but that was before technology became sufficiently advanced."