Can AI be Trusted with Surveillance Tasks?

China's war on jaywalking went to the next level last spring when AI-based facial recognition systems were integrated into some crosswalks, to punish jaywalkers by squirting them with water, sending them texts warning them about legal consequences of jaywalking, and/or publicly shaming them by displaying their pictures and names on large digital billboards.

Last week, this system entered a new and exciting failure mode when a traffic-cam in the port city of Ningbo captured a face displayed on the side of a passing bus, correctly identified it as belonging to Dong Mingzhu, CEO of Chinese AC giant Gree Electric Appliances, and then plastered Ms Dong's face all over a giant billboard, falsely accusing her of jaywalking. more

Facial Recognition Technology – Not Ready for PRIME Time

Amazon.com’s facial recognition tools incorrectly identified Rep. John Lewis (D-Ga.) and 27 other members of Congress as people arrested for a crime during a test commissioned by the American Civil Liberties Union of Northern California, the watchdog said Thursday...Amazon’s so-called Rekognition technology — already in use at law-enforcement agencies in Oregon and Orlando — is hampered by inaccuracies... more

Security Scrapbook Flashback to 2008.

You Are Never a Stranger in Our City - Facial Recognition Street Cams

Nvidia has partnered with AI developer AnyVision to create facial recognition technology for "smart cities" around the world. The two companies will work to install automatic facial recognition into CCTV (closed-circuit television) surveillance cameras.

AnyVision claims the technology enables cameras that can continuously scan for faces 24/7, automatically identifying and tracking individuals within a large crowd with 99% accuracy.

Algorithms working with human monitors can then compare the faces identified against a database of known terrorists or criminals.

The company also says it's committed to protecting the personal data that CCTV cameras collect.

So... is this technology terrifying, and possibly everything Orwell warned us about? Absolutely.

But it could also save thousands of lives. The technology could be useful not only for catching at-large criminals, but also for quickly identifying suspects, and tracking down individuals who have gone missing. more

You may recall, this has been tried before and its comeback was predicted here in 2008.

Facial Recognition Glasses Used by Police to Identify Suspects

Chinese police are using dark sunglasses equipped with facial recognition technology to spot criminal suspects.

The glasses, which are being worn by police at a busy train station ahead of the Chinese New Year travel rush, are linked to a central database which contains details of criminal records.

Wearing the technology, police can almost instantly view an individual's personal details, including name, ethnicity, gender and address.

The scene would not look out of place in an episode of science fiction television drama Black Mirror, which often depicts dark scenarios of humans being overcome by technology.

China is deploying new technologies to monitor people in ways that would unnerve many in the West. more

33 Years Late, or You'll Never be a Stranger Here

China has been building what it calls "the world's biggest camera surveillance network".

Across the country, 170 million CCTV cameras are already in place and an estimated 400 million new ones will be installed in the next three years.

Many of the cameras are fitted with artificial intelligence, including facial recognition technology. The BBC's John Sudworth has been given rare access to one of the new hi-tech police control rooms. excellent video demo

Anti-Surveillance Sunglasses – Q Would Be Proud

via... digitaltrends.com
...there’s a new set of spectacles on Kickstarter that might help you bamboozle even the most sophisticated facial recognition tech.

The Eko shades, as they’re called, are rimmed with a type of retro-reflective material that bounces light back to exactly where it came from. Most surfaces reflect light by diffusing or scattering it in all directions, but this material is specially designed to reflect light back at the exact same angle as it arrived.

If caught in flash photography, retro-reflective material will send most of the light back to the camera’s sensor. This will put the dynamic range of the camera’s sensor to the test, and likely result in an image that’s underexposed for everything but the rims of your glasses.

Of course, this won’t help much for any camera that doesn’t require a flash, but it’s still a pretty interesting concept. more

...and the DIY hat to go with them!
1937 prototype anti-mind control device.
(Ok, who said, "Too late. She has already lost hers.")

Odd-Ball - Anti Facial Recognition to Debut at Sundance Film Festival

HyperFace is a new kind of camouflage that aims to reduce the confidence score of facial detection and recognition by providing false faces that distract computer vision algorithms...

HyperFace will launch as a textile print at Sundance Film Festival on January 16, 2017.

Prototype

HyperFace works by providing maximally activated false faces based on ideal algorithmic representations of a human face. These maximal activations are targeted for specific algorithms. The prototype is specific to OpenCV’s default frontalface profile. Other patterns target convolutional nueral networks and HoG/SVM detectors... HyperFace reduces the confidence score of the true face (figure) by redirecting more attention to the nearby false face regions (ground).

Conceptually, HyperFace recognizes that completely concealing a face to facial detection algorithms remains a technical and aesthetic challenge. Instead of seeking computer vision anonymity through minimizing the confidence score of a true face, HyperFace offers a higher confidence score for a nearby false face by exploiting a common algorithmic preference for the highest confidence facial region.

In other words, if a computer vision algorithm is expecting a face, give it what it wants. more

Russian Hi-Tech Spy Devices Under Attack Over Privacy Fears

New Russian technologies, including phone call interception and a facial recognition app, have stirred a fierce debate about privacy and data monitoring. 
 
Infowatch, a Moscow-based IT security company managed by businesswoman Natalya Kasperskaya, found itself in hot water last month after it revealed it had invented a system that companies can use to intercept employees' mobile phone conversations...

The goal behind phone call interception, Kasperskaya said, is to provide large businesses with a tool to prevent information leaks, including companies whose success depends on protecting corporate secrets. more

Yet More Invisibility Eyeglasses

This year, AVG will reveal a set of concept invisibility glasses at Pepcom in Barcelona before Mobile World Congress. 

What are invisibility glasses?

Developed by AVG Innovation Labs, the glasses help protect your visual identity in the digital age.

Through a mixture of technology and specialist materials, privacy wearables such as invisibility glasses can make it difficult for cameras or other facial recognition technologies to get a clear view of your identity.

...there are generally two different methods of combating unwanted facial recognition:

Infrared Light
The idea is to place infrared LEDs inserted around the eyes and the nose areas. Since the infrared lights are completely invisible to human eyes, they are only detectable by cameras which are sensitive to the wavelengths of these LEDs. They claim to break face detection when the lights are on.

In this example we show how infrared can be used to avoid Facebook’s facial recognition technology.

Retro-reflective Materials
These specialist materials help maintain your privacy at the moment that the image is actually taken.

PS - This is a proof-of-concept project. Not for sale. However, you can make your own.

Other glasses.

Note: Many cell phone cameras have infrared cutoff filters built into their lenses... and you can bet law enforcement facial recognition systems do too.

Proof the Surveillance Society is Making us Crazy - CV Dazzle

This is how bad things are getting...

"The NSA made me slather my face in make-up... I had slathered the paint on my face in order to hide from computers. The patterns in which I applied the paint were important: To the pixel-calculating machinations of facial recognition algorithms, they transformed my face into a mess of unremarkable pixels. In the computer’s vision, my face caused a momentary burst of confusion. That’s why the patterns are called computer vision dazzle (or CV dazzle). When it works, CV dazzle keeps facial-recognition algorithms from seeing a face...

...more unexpected was what CV dazzle taught me about the physical world. It reminded me of another tech experiment I’d undertaken: 

My phone’s Reminders app can tie a message to a specific place, it triggers an alert tone every time a user comes within 500 feet. I’d tried tying these reminders to a different kind of location—the 176 embassies and diplomatic missions in Washington, D.C. Whenever I got within a couple hundred feet of one, my phone sent me a little ping: “Iceland.” “Thailand.” “Equitorial New Guinea.”...

...here is the essence of CV dazzle’s strangeness: The very thing that makes you invisible to computers makes you glaringly obvious to other humans."  (more) (official site cvdazzle.com)


Blank Reg would have loved this.

Mannequin Spies - Will Dummy Shoppers Revolt?

An Italian firm selling mannequins that secretly monitor the age, race and gender of customers using facial recognition software has come under fire from privacy groups. The information logged by the dummies is then used to implement more effective marketing strategies by stores in the US and Europe.

And the manufacturer now plans to add audio recording to the dummies' capabilities, listening in on customers' discussions about their clients' products.

Click to enlarge

The mannequins, known as "EyeSee" are manufactured by Italian company Almax and retail for £3,200 each.

Privacy campaigners agree, describing the technology as "creepy" and "totally disproportionate."

Emma Carr, deputy director of campaign group Big Brother Watch, told the Daily Mail newspaper: "The use of covert surveillance technology by shops, in order to provide a personalised service, seems totally disproportionate.

"The fact that the cameras are hidden suggests that shops are fully aware that many customers would object to this kind of monitoring. 

"Keeping cameras hidden in a mannequin is nothing short of creepy." (more)

As Seen on Predicted on TV in 1960!

(YouTube)

 

The New York City Police Department now has "The most advanced and technological counter-terrorism bureau that anyone has ever seen."

NY - A new crime-tracking system designed jointly by the New York Police Department and Microsoft Corp. will pool existing data from cameras, 911 calls and other technologies to provide crime fighters with a comprehensive view of threats and criminal activity, as well as provide the city with a new revenue source.

The Domain Awareness System will be able to map suspects' movements and provide NYPD investigators and analysts with real-time crime alerts.

...the system will allow NYPD personnel to track a suspect's car, and find out where it's been located in the past days or weeks synthesizing archived video footage and license plate reader data. Other potential uses include mapping criminal history geospatially and chronologically to reveal patterns, and the ability to instantly see suspect arrest records, 911 calls associated with the suspect and related crimes occurring in the area. (more) (60 Minutes video) 
This afternoon the NYPD debuted their "all-seeing" Domain Awareness System, which syncs the city's 3,000 closed circuit camera feeds in Lower Manhattan, Midtown, and near bridges and tunnels with arrest records, 911 calls, license plate recognition technology, and even radiation detectors. Mayor Bloomberg dismissed concerns that this represented the most glaring example of Big Brother-style policing. "What you're seeing is what the private sector has used for a long time," Bloomberg said. "If you walk around with a cell phone, the cell phone company knows where you are…We're not your mom and pop's police department anymore."

NYPD Commissioner Ray Kelly stated that the system, which is currently operational out of the department's Lower Manhattan Security Commission HQ, was developed with a "state of the art privacy policy" and "working with the privacy community," but did not offer specifics. DAS does not have facial recognition technology at this time, but "it's something that's very close to being developed," the mayor said.  

The system was developed with Microsoft and paid for by the city for $30 to $40 million, and has already been in use for six months. The feeds compiled by the system are kept for thirty days, then erased.

The City will receive 30% on the profits Microsoft will make selling it to other cities, although Mayor Bloomberg declined to say if that money would go back into the NYPD. "Maybe we'll even make a few bucks." (more)

New CCTV Scans 36 Million Faces for a Match... in one second!

There were several news stories late last week about a new surveillance system by Hitachi Kokusai Electric that the company claims is able to capture a person's face and, in one second, scan some 36 million facial images stored in its database to see whether it can find a match. According to this story at Digital Trends:

"Now, here's my plan..."

"Hitachi’s software is able to recognize a face with up to 30 degrees of deviation turned vertically and horizontally away from the camera, and requires faces to fill at least 40 pixels by 40 pixels for accurate recognition. Any image, whether captured on a mobile phone, handheld camera, or a video still, can be uploaded and searched against its database for matches." 

The company states in a video posted at DigInfoTV that it thinks the system is "suitable for customers that have a relatively large-scale surveillance system, such as railways, power companies, law enforcement, and large stores."

Over time, I suspect that the technology will be reduced in price to be "suitable" for just about anyone with a surveillance system. (more)

FutureWatch: Your Next TV May Watch You

via the HD Guru...

Artist's conception. Not really Samsung.

Samsung’s 2012 top-of-the-line plasmas and LED HDTVs offer new features never before available within a television including a built-in, internally wired HD camera, twin microphones, face tracking and speech recognition.  

While these features give you unprecedented control over an HDTV, the devices themselves, more similar than ever to a personal computer, may allow hackers or even Samsung to see and hear you and your family, and collect extremely personal data.

And unlike other TVs, which have cameras and microphones as add-on accessories connected by a single, easily removable USB cable, you can’t just unplug these sensors.

Privacy concerns

We began to wonder exactly what data Samsung collects from its new “eyes and ears” and how it and other companies intend use it, which raises the following questions:

* Can Samsung or Samsung-authorized companies watch you watching your Samsung TV? 

* Do the televisions send a user ID or the TV’s serial number to the Samsung cloud whenever it has an Internet connection? 

* Does Samsung cross reference a user ID or facial scan to your warranty registration information, such as name, address etc.? 

* Can a person or company listen to you, at will, via the microphone and Internet connection? 

* Does Samsung’s cloud store all this information? How secure is this extremely personal data? 

* Can a hacker intercept this data or view you via the built in camera? 

* Can a third-party app program do any of the above? 

* Exactly what information does the TV send to Samsung or other parties? 

* Does Samsung intend to sell data collected by its Smart TV owners, such as who, what and when one is viewing? (more)

Could Facial Recognition Become the Next Emergency Broadcast System

The Emergency Broadcast System (EBS), a communications system which allows the government to commandeer radio and television broadcasting outlets to distribute emergency messages quickly, was tested this past week. The EBS started in 1963, and was preceded by a similar service called, CONELRAD. Electronic eminent domain has been around a long time. Hold that thought.

Today, The New York Times reports Face Recognition Makes the Leap From Sci-Fi.

"SceneTap, a new app for smart phones, uses cameras with facial detection software to scout bar scenes. Without identifying specific bar patrons, it posts information like the average age of a crowd and the ratio of men to women, helping bar-hoppers decide where to go. More than 50 bars in Chicago participate... The spread of such technology — essentially, the democratization of surveillance — may herald the end of anonymity. 

Those endeavors pale next to the photo-tagging suggestion tool introduced by Facebook this year... “Millions of people are using it to add hundreds of millions of tags,” says Simon Axten, a Facebook spokesman. Other well-known programs like Picasa, the photo editing software from Google, and third-party apps like PhotoTagger, from face.com, work similarly. 

And this technology is spreading. Immersive Labs, a company in Manhattan, has developed software for digital billboards using cameras to gauge the age range, sex and attention level of a passer-by.

Using off-the-shelf facial recognition software, researchers at Carnegie Mellon University were recently able to identify about a third of college students who had volunteered to be photographed for a study — just by comparing photos of those anonymous students to images publicly available on Facebook."

Have you connected the dots yet? 

Here is another clue...  

CALEA, a law passed in 1994, "To amend title 18, United States Code, to make clear a telecommunications carrier's duty to cooperate in the interception of communications for Law Enforcement purposes, and for other purposes." Telecommunications yet another type of mass communications device which may now be commandeered by government.

Internet connected, facial recognition systems are rapidly becoming mainstream mass communications technology, just like radio, TV and telephones. It only makes sense that this too will be commandeered. The question is, will it be commandeered like EBS to broadcast emergency messages, or will it be commandeered like CALEA to be used for surveillance? Both, perhaps?

So far, the benefits of letting government commandeer mass communications (verses the abuse potential) make the gambit worthwhile. For this, we thank our legal system. It is time for them to walk the high wire again. Please, get us through this technical conundrum with grace and balance one more time.

The noose tightens... "You can run, but you can't hide." 

Should this all come to pass (it will), there may be some interesting social outcomes. Just as mass communications pulls society closer together, mass surveillance capabilities like CALEA, license plate readers, and the multitude of facial recognition surveillance systems may push people apart. Imagine a world where the density of: commercial video billboards and kiosks; business surveillance cameras; and government street/toll booth cameras, in urban areas, squeezes criminals into the suburbs and beyond.

How best to take advantage of the changes in our brave new world? 

I have a career tip for you.

~ Kevin

"You're only a stranger here once!" ~Tampa, FL

Do you recall my prediction about Tampa?
FutureWatch (September 2008) - Although facial recognition and tracking didn't catch on the first go-around (the Tampa, Florida experiment), it is ripe for a come-back. 5 years from now, this will be commonplace – along with automatic license plate readers and motion-intention evaluators.

August 2003 - Tampa police have scrapped their controversial security camera system that scanned city streets for criminals, citing its failure over two years to recognize anyone wanted by authorities.

History...
July 2001 - The Tampa City Council took a fully-informed look at Ybor City's controversial high-tech face-scanning software. When the dust settled, the council split down the middle with a 3-3 vote on whether or not to do away with the face-scanning software.

Fast Forward... 2011 - via National Motorists Association...

The request reads like a shopping list for a counter-terrorism strike: low-light cameras to identify people and vehicles at 100 meters, helmet-mounted cameras, cameras for "use around high-risk activities" and cameras that can read license plates across three lanes of traffic.

In reality, it’s part of a plan proposed by Tampa city officials to provide security for next year’s Republican National Convention. Funds to buy or lease the gear are expected to come from federal taxpayers in the form of a $55 million congressional appropriation.

The surveillance will target convention protestors (as many as 10,000, according to convention organizers), but, given the sweeping nature of the plan, many bystanders and motorists are likely to be ensnared as well.

And while police officials admit they may not get all 238 cameras on the original request, critics are already reacting. A spokesperson for the American Civil Liberties Union of Florida likens the approach to "hitting a gnat with a sledgehammer." (To be fair, officials canceled a request for two aerial surveillance drones due to cost concerns.)

FutureWatch - Drones are already in some state and local police toy chests. Tampa will eventually get one, too.

The End of Anonymous On-Line Dating Has Arrived

Science fiction writers have long imagined a future in which facial recognition technology makes anonymity in public obsolete. A research study at Carnegie Mellon University suggests that this Minority Report future has already arrived, thanks to facial recognition products now commercially available, combined with the 750-million-person identity database called Facebook.

A CMU research team led by associate professor Alessandro Acquisti took candid photos of 93 random students on campus using a $35 webcam. Within seconds the researchers were able to determine the identities of a third of their photogenic guinea pigs, using off-the-shelf facial recognition software from PittPatt, a software company recently acquired by Google, and publicly available profile photos from Facebook. The researchers had an even higher rate of success using the same technology to identify more than 100,000 Pittsburgh singles with otherwise pseudonymous accounts on a dating site, adding yet more complexity to the world of online dating. (more)

FutureWatch Prediction Comes True - Tampa Redux

Tampa - "You are only a stranger here once." 

August 2011 - The Tampa Police Department plans to circle its downtown area with surveillance cameras prior to the 2012 Republican National Convention, but Assistant Police Chief Mark Hamlin says the locations will not be disclosed in order to keep potential troublemakers from gaining a tactical advantage. A local political activist has publicly stated he is troubled by this decision.

September 2008 - FutureWatch - Although facial recognition and tracking didn't catch on the first go-around (the Tampa, Florida experiment), it is ripe for a come-back. 5 years from now, this will be commonplace – along with automatic license plate readers and motion-intention evaluators.

August 2003 - It is with sadness we note the demise of our favorite city motto... Tampa - "You're only a stranger here once." Tampa police have scrapped their controversial security camera system that scanned city streets for criminals, citing its failure over two years to recognize anyone wanted by authorities.

History - July 2001 - The Tampa City Council took a fully-informed look at Ybor City's controversial high-tech face-scanning software. When the dust settled, the council split down the middle with a 3-3 vote on whether or not to do away with the face-scanning software.

Tampa Police Radio Room c.1920's

(How Antonio Prohías might see it.)

Oh, one more thing...

One security feature I would like to see on my future cell phone is the option of not using a password.

Think of this... all business-level cell phones have camera capability; all have (or could easily be designed to have) touch screen capability; and of course a microphone. The next logical step is adding facial, fingerprint or voice recognition to replace the access PIN code. 

In addition to the security benefit, it would sure make using the phone while driving safer. (Just kidding. I would never do that. Well... not often, anyway.) ~Kevin

Can you zoom in on that license plate?

The recent video enhancement post sparked additional interest in the subject. Upon doing some more research, it became clear that enhancing is only part of the picture. (Geez, I am starting to sound like Mr. Peabody.) 

Even more important is investing in technology which works in the first place. For example, the video surveillance system required in a small store is not what works best for surveying a parking lot.

The following explanation at Jim Hoerricks' blog is really insightful and educational...

Can you zoom in on that license plate? Can you zoom in on his face? Why can't you make this image clearer. These are questions that many forensic video analysts deal with on a daily basis. To illustrate the answer to these questions, let's consider this image:

This Monitoring image comes from an Avigilon system's 16 megapixel camera.

With the above image in mind, let's consider the outstanding advice from the UK's Home Office Scientific Development Branch:

To judge the quality of images that will be necessary, you will need to take into account the purpose for which CCTV is used and the level of quality that will be necessary to achieve the purpose. The Home Office Scientific Development Branch (HOSDB) recommends identifying the needs of a CCTV system by using four categories:

1. Monitoring: to watch the flow of traffic or the movement of people where you do not need to pick out individual figures.
2. Detecting: to detect the presence of a person in the image, without needing to see their face.
3. Recognising: to recognise somebody you know, or determine that somebody is not known to you.
4. Identifying: to record high quality facial images which can be used in court to prove someone’s identity beyond reasonable doubt.

Take a look at the picture above. Can you make out faces or license plates? No. So, if your purpose is to identify these items, that picture view just won't do. 

With Photoshop, we can zoom in. Let's see what happens.

This Detecting view comes from zooming in a 16 megapixel image.

We can now see the people and vehicles more clearly. We can gain a general awareness of types of vehicles. We are closer to identifying them, but we don't have anything yet that will help us to prove identity beyond a reasonable doubt. We can zoom in further to see if we can Recognise anyone.

This Recognising view comes from zooming in a 16 megapixel image.

 At this level of magnification, someone who knows this person or car should be able to say, "that's him" or "that's not the man/car you are looking for." At this point, we can begin to accurately describe the individual in the scene. Let's zoom in some more to see if we can positively identify this man and the car.

This Identifying view comes from zooming in a 16 megapixel image.

With a 16 megapixel image, we can even zoom in closer than this image. But notice what's happened along the way. We've sacrificed field of view for detail. No longer are we looking at the whole parking lot (monitoring). We can't see the other people who are walking around (detecting). In order to identify this individual, we've had to zoom in so far that we've excluded much of this scene from our view. A CCTV Installer might position a camera/lens combination at a choke point specifically to get facial recognition whilst installing other cameras around the area to monitor and detect movement of unauthorised persons.

But what about real life. The good folks at Avigilon have a small piece of a very large market. What if that same monitoring image was only 4CIF or 2CIF. Could we still zoom in and identify the individuals in the scene? You be the judge.

The image from Avigilon contains 15,824,256 pixels and can come in a lossless RAW format.

This image contains only 426,400 pixels and is compressed. 

With an almost 97% reduction in the amount of available pixels and the additional compression, the results speak for themselves.

The result of zooming in on a low pixel count image.

So, the answer to "can you zoom in on that license plate" is ... it depends on the quality of the image and the number of available pixels. ~Jim Hoerricks - author of Forensic Photoshop, a comprehensive imaging workflow for forensic professionals