How to Beat Facial Recognition

Can you commit a crime and go on the run for years without being caught today? This is what the experts say…

Source: CV Dazzle

ast week, I watched a movie called “The Old Man & the Gun”, based on the true story of Forrest Tucker, a 60-year man who escaped from prison and conducted some 60 bank robberies while on the run for more than three years before being caught. This was back in 1980.

I started to wonder if all that would still be possible today, given all the CCTV and facial recognition software in use. Could anyone still go on the run for so long without being caught?

1.8 billion faces in three seconds

I had a client who is one of the top three leaders in facial recognition technology in China. Their software reportedly helped to identify and capture hundreds of criminals in just three years, including a murder suspect who had been on the run for 16 years.

There are an estimated 176 million CCTV cameras in China today, and my client’s software claims to be able to screen 1.8 billion pictures in three seconds. They also claimed to be able to handle aging or partially hidden faces with their face matching algorithms.

Out of intellectual curiosity I started to wonder if there was a reliable way to beat facial recognition technology.

To my surprise this was what I found…

Star Wars had the solution?

The solutions found on the internet ranged from the serious to the hilarious. Among the latter there was a dude who proposed wearing a Darth Vader mask or tilting your head at a 15 degree angle all the time when you are walking on the streets.

Source: Aram Bartholl on Youtube

Not so practical.

Perhaps a method that would draw less attention would be wearing a hoodie like the Jedi’s in Star Wars. But be warned, many countries have laws against wearing masks or clothing that intentionally covers your face. (In the U.S. such laws date back to fighting the Ku Klux Klan…)

What about makeup?

Mike Harvey, a designer, created a brand called ‘CV Dazzle’ out of a series of hair and makeup looks targeted at beating facial recognition software.

But for CV Dazzle’s designs to work, one had to look like you were going to a punk rock concert.

Source: CV Dazzle

An experiment by Mashable showed that even if you put on heavy makeup, the iPhone X’s face recognition software would still successfully unlock your phone upon scanning. That’s because it uses infrared light images to determine contours on the face rather than matching colors or flat images.

Heavy makeup isn’t enough to beat face recognition. Source: Mashable on Youtube

However, a style of face painting called ‘Juggalo’ makeup — that makes you look like black metal musicians— is heavy enough to fool the software.

The article attributed this to the fact that “applying black paint below the mouth, but above the chin…makes facial recognition vulnerable to misidentifying the placement of the jaw”.

And now the reality… caps, scarves and glasses wont’ help

Partially obscuring your face isn’t going to help with today’s technology.

Back in September 2017, a researcher at Cambridge University had already developed a software that could accurately identify individuals wearing caps, scarves and glasses more than half the time. The technology reportedly also negated hair and makeup solutions like the kind CV Dazzle has.

While this prototype system is based on facial feature extrapolation using machine learning algorithms, Apple has confirmed that its infrared light technology on iPhone X can work even if users are wearing sunglasses.

Moving on from low tech to high tech

So if facial recognition is becoming high tech enough to recognize partially obscured faces, and it is illegal to completely mask your face, so what then?

Perhaps it makes more sense to fight high tech with high tech.

Various researchers from east to west have devised techie headgear to fool facial recognition systems.

Professor Isao Echizen from Japan has gone thru’ at least three iterations in his quest to invent a ‘privacy visor’. In his first prototype, he used small flashing LED lights to confuse facial recognition technology.

The second one involved using “unique patterns and angles” on the lenses to reflect, refract or absorb light. Wall Street Journal reported that it worked 90% of the time with smartphone cameras. But while you could walk around easily enough in it, the official advice was not to wear it while driving or even riding a bicycle…

Source: engadget

That was back in 2015. Two years later Prof Echizen came back to the market with an improved version that looked more like a day-to-day sunglass. All the user had to do to avoid detection was to tilt the lenses up by 10 degrees (to reflect light around the eyes away) when required.

A team of cross-border researchers have also tried the LED solutions on a baseball cap. It worked 70% of the time.

“In 2018, researchers from Fudan University in China, the Chinese University of Hong Kong, Indiana University, and Alibaba Inc., used an array of tiny infrared LEDs wired to the inside of a baseball cap to project dots of light onto the wearer’s face. These dots were invisible to the human eye, but confused the computer vision and made the face unidentifiable”

How to hack your face to dodge the rise of facial recognition tech, Wired Magazine, February 1, 2019

A company from Chicago called Reflectacles has also invented glasses that blocks out infrared light. Their latest Kickstarter campaign has raised about US$34k from 193 backers and the product is expected to ship in April 2020.

Prevention is better than cure

Interestingly, while most solutions focused on not being recognized, one researcher looked at how to prevent yourself from becoming part of the database for matching.

Joey Bose, a computer engineering student at the University of Toronto, invented an app that can subtly modify elements of a photo before it is uploaded onto the internet.

Source: Fortune Magazine

“The photos don’t look any different to the naked eye, but the hidden features thwart detection systems.”

Here’s a New Way to Trick Facial Recognition

Unfortunately his technology helps you only if you haven’t spent the last decade flaunting your handsome face or svelte figure on Instagram or Facebook, which many systems trawl to build their facial recognition databases.

Bose chose to turn down interest from venture capital and instead pursue research and his PhD.

The solitary conclusion

As I read more and more about facial recognition tech and counter-measures, it seemed more and more obvious that it was hard to completely avoid detection unless one was prepared to live in a cave or hide in your home all day.

As Wired magazine puts it, “Any method for evading recognition is at best a stopgap solution, particularly when FR is combined with other forms of biometric identification.”

Technology is always a double-edged sword. It could be used for good or evil depending on the wielder.

As the debate between privacy vs security rages on, I think the important question isn’t about how much or how little of surveillance to impose, but to undertake a realistic study of just how effective is modern surveillance technology in crime prevention.

While facial recognition technology can probably result in a modern day ‘Forrest Tucker’ being arrested a lot faster than three years, can it effectively deter someone from becoming a serial robber, gunman or rapist?

Have we truly — with technology — gone from detection… to prevention?

Only then can society objectively rationalize and weigh the cost of privacy versus the benefits of security.

Investor | Entrepreneur | Thinker 🔗

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store