How to Beat Facial Recognition Software

February 22, 2021


  • Already In Use
  • How It Works
  • Hiding in Plain Sight
  • How You Should Avoid Detection

Imagine a world where everywhere you go, you are being watched, scrutinized, penalized, or rewarded for your behavior.  In some countries, this is is their reality.  If you run a temperature, a camera connected to some software locks the door and bars your entrance.  You step out into a walkway but catch yourself almost walking against the light, so you step back onto the sidewalk.  It’s too late, though. Your jaywalking ticket is already in the mail to you.  Imagine the marketing agencies knowing when you went into a retail store, how long you lingered, what you bought, how you paid for the items, and how they are used.  

It’s not hard to imagine because the capabilities of the software, hardware, facial detection, and facial recognition software have grown tremendously over the last decade and the aggregation of data will explode over the next few years. Such systems are already in place and are already making decisions about how you can live your life.  With each passing month, the systems become more integrated with each other.  As your retail purchases database and spending habits sync up with your online search history and your income and demographics information, tying your face to the data is the final piece of the puzzle.  Once that piece is ubiquitous in public places, anything from targeted advertisements to all-point-bulletins is possible.  Doors can be locked in your face.  Tickets can be issued.  Rewards can be given, or fines can be incurred.  Your smartphone, the most prominent tracking device you willingly carry with you, is nothing compared to the exact signature of your face.  In this video, we will look at methods for keeping your identity safe from the obtrusive and evasive prying cameras of the state, what works and what doesn’t work, and what you should be doing now to protect your facial biometrics.

Already in Use

Our faces are being scanned daily and recorded and saved on databases without our consent.  In theory, if you are not a wanted person, as they say, you shouldn’t be concerned, right?  Wrong.  The fact is that the applications that are currently scanning faces are numerous.  The places where your face is being scanned are numerous.  The laws that protect your privacy are few and ineffective.  If a law is passed that states, for instance, that facial scanning cannot be done in real-time, the programmers just delay the live stream feed for a few seconds and proceed normally with their surveillance.  There is no entity to report these activities to, so Russian app designers have even developed seemingly benign facial morphing apps that amusingly distort one’s appearance– aging a person, making them younger, showing them as another gender, and so forth.  In the terms and conditions, the users agree to sign over the rights to any images they submit.  So, users are signing over to a Russian company the image of their own face.

In China, a celebrity whose face was on an advertisement on a city bus received a jaywalking ticket because her face was detected in the image as the bus went by.  Facial recognition is increasingly being used to scan and monitor crowds in public transportation systems, stores, banks, stadiums, and even public streets and parks.  Facial detection and facial recognition software are a passive means to observe large swaths of the population and alert on any key features the end-user desires.  If you only want to find a person meeting a specific description, you could plug that person’s image in for the software to compare to all incoming live images.  For instance, the software could be programmed to find any faces of a certain height, with specific hair color, and it would trigger on those faces in a crowd.

How it Works

The first part of facial recognition is facial detection.  Here, the electronic image or data stream is scrutinized to determine if a face exists.  The second part is actual facial recognition.  A facial recognition system is a technology capable of matching a human face from a digital image or a video frame against a database of faces.  Facial recognition systems have been deployed in advanced human-computer interaction, video surveillance, and automatic indexing of images.  The systems are easy to build and easy to deploy.  They are not, typically, actively monitored in real-time but set off a trigger or an alert to a human operative who then acts upon the information the system provides.  It doesn’t have to be a human interaction, though.  Other software can be designed to initiate actions based upon the triggered face.  

Imagine all the walk signs and street lights set against you.  The triggering of a light is a programmed piece of software.  It’s a set series of inputs resulting in a set sequence of outputs.  I’m not saying that the next time you catch every red light, there’s some larger plot against you, but I am not saying it’s an impossibility either.  More likely is your face triggering an advertisement or a print out coupon on your receipt at checkout.  While none of these are too invasive, we are at the early stages of wide-scale integration and facial recognition deployment.  Therefore, you should be aware of how your face is captured and how it is a unique identifier of you.

Facial recognition software identifies key points on a face: the distance between pupils, the outer end of the eye or outer canthus, the inner point of the eyes or inner canthus, the center point between eyes or glabella, the tip of the nose, the philtrum, the nares, the chin, the points around the lips, and many more spots.  The more points the software can identify and measure, the more accurate its identification will be.  The uniqueness of our faces is as precise an identifier as our fingerprints.  Our brains do the same measurements and identification, but we just don’t realize that they are doing it.  And, just like our brains, the software has a baseline.  It compares those points to a set of matching parameters or records the information for further comparisons.

How often is it occurring?  Well, most systems were CCTV or closed circuit.  This means they are not intentionally out there for anyone to view; however, most video capturing systems today are hackable and can be accessed over the web no matter how locked down their manufacturers say the software is.  How many CCTV domes do you see around banks, ATMs, parking structures, public parks, in front of doorways, video doorbells, and so forth?  Pause for a moment and add up all the cameras in a fifty-foot radius around you: every laptop, every smartphone, every store, every dashcam, and every street camera, every video doorbell.  Imagine that they are all linked together.  Does it provide a comprehensive view of an entire area?  Likely, they are not all linked together at any given moment, but it only takes one of the streams from one of the cameras connected to software programmed to be used for facial recognition to identify you.  It only takes two or more of those streams to predict your movements.  Thankfully, these numerous systems are not all linked together yet, but what can you do now to protect yourself in the future?

You can’t avoid being captured in the camera’s range because you can’t become invisible.  However, you can reduce your chances of detection and confuse cameras and recognition software in many ways.  Disrupting the facial detection and facial recognition through concealment, therefore depriving the data needed or disguising where the data picked up is flawed, are highly effective methods to remain unseen. 

Hiding in Plain Sight

Pattern recognition is another method by which facial recognition software detects faces.  The human face has dark spots and light spots.  Those all match a similar pattern.  When those are reversed, a person can sometimes elude detection.  This isn’t as practical of a method, however, because you would stand out if you had makeup on that made your chin, cheeks, and forehead exceedingly dark while your eye areas were extra bright.

In all seriousness, you could also wear the printed out face of another person.  Through these methods, you may avoid detection by cameras but not by any humans.  You would stand out to the average viewer.  A more subtle way would be to have a picture of people on your clothes.  This can sometimes overwhelm the algorithms used to interpret the images.  While it is busy analyzing the image, you are sometimes overlooked.  This method, too, is not 100% effective, and some technologies will still be able to pull your face out of the mix of other images on your clothes some percentage of the time.  Wearing large faces on your clothing or loud patterns that appear to facial recognition as faces may be a loud fashion choice and may result in you rising to the attention of other humans.

Anti-facial recognition clothing, face masks, sunglasses can all work to thwart facial detection and recognition systems.  Some methods are more obviously seeking to thwart these systems and may draw attention by way of their audacity.  Others, like reflective acetate tapes and films applied to glasses or hats, can spoof infrared systems by not allowing them to achieve an accurate read on a person’s face.  There are also reflective paint sprays on the market that will reflect infrared light to confuse cameras.  These will also make you stand out brightly at night or under blacklights.  Essentially, they create a blind spot by overwhelming the person’s location in the infrared camera’s field of vision.  Infrared and ultraviolet blocking films for windows can be applied to glasses to hide features behind clear glass.  Blocking films, however, will be the opposite of reflective acetates and paints. They will darken the area to the cameras, thereby concealing the features of the eyes.  Eyes are ordinarily dark to facial detection software, so your face will be detected, but the features will not be recognized.

Physical tricks like tilting the head, concealing the bridge of the nose, placing a dark gloved hand on your face, or simply looking down can enhance your concealment from facial detection and recognition software.  These methods are not the most reliable and do not work all the time.  Also, as detection systems get more sophisticated, these tricks are being coded into the detection.  At some point, one of these tactics may actually become a signaling flag.  Fortunately, cameras need a wide-angle and a large area to scan.  To accomplish this expansive view, cameras have to be positioned above the average height of a human and pointed toward crowded areas.  Knowing this, merely looking down while wearing a wide-brimmed hat like a baseball cap would be enough to conceal precise details.  Though thermal imagery can see through sunglasses, details are not rendered clearly when wearing glasses.  

Infrared light is used in most surveillance systems because it is invisible to the human eye and can illuminate an image for the camera in extremely low light situations.  Technically, it is electromagnetic radiation with wavelengths that are longer than visible light.  Most of the thermal radiation emitted by objects near room temperature is in the infrared spectrum.  Because of this, lining jackets, hats, or the bridge and chin area of a facial mask with strips of a typical mylar emergency blanket can alter the heat signature of a person.  Reducing heat signatures can throw off surveillance systems, as they zero in on more prominent sources.  The mylar portion will block the infrared signature of what is behind it.

While no single one of these outlined methods is guaranteed wholly effective, some combination of a little bit of each will be extremely useful.

How You Should Avoid Detection

Though there are many ways to try and trick cameras and facial detection and recognition software, you can adopt a casual combination of a few of them to become invisible to most passively observing systems.  If facial recognition systems are specifically targeting you, naturally, the methods you adopt will need to be more extreme.  You very likely don’t want to go about your everyday life wearing a prosthetic nose to extend your nose’s tip, so I do not suggest that extreme for everyday living.  I suggest that you adopt just a few methods to casually practice daily to help you fly under the radar of detection systems.  As they become more advanced, you will be able to stay out of their databases.

Here are a few methods you should adopt if facial recognition is a concern for you.  First, get in the habit of wearing a hat and glasses.  These will conceal your face’s finer points, which will prevent a camera from receiving enough data to be compiled by facial recognition software.  Using ultraviolet and infrared-reflective or blocking materials on your hat and glasses will further confuse these systems.  If you don’t like wearing a hat or you would be out of place doing so, realize that facial detection and recognition software relies on symmetry to identify faces.  Even wearing your hair over one side of your face can throw off this symmetry.

Transparent glasses with blocking or reflective films will obfuscate the endpoints of the eyes and the centers of the iris and pupils.  Without that data, the software has a more difficult time achieving certainty.  Without certainty, the software is less likely to alert any human operators because the programmed parameters haven’t been met.  Sunglasses are useful but are not reliable when the sun isn’t shining, as they will make you stand out from the crowd.  If that doesn’t matter and you are only concerned with the recognition software identifying you, sunglasses will be more effective than clearer glasses.

Second, wear a mask or a scarf if the weather permits.  A few years back, wearing a mask would make you a focal point.  Now, not wearing a mask is more likely to draw the observer’s gaze.  A mask will conceal all the points on your nose, your mouth, and your chin.  If there isn’t a clear image of your eyes, there simply isn’t enough data for any software in the world to compile with certainty your identity.  Wearing a patterned mask designed to confuse cameras can also reduce the likelihood of detection.  Protect your face’s identification points like you protect your driver’s license if you hope to remain unseen.

Finally, make sure that your devices aren’t feeding into a more extensive system.  Do your kids sometimes have Zoom calls with grandma?  Make sure your camera on your laptop is covered when not in use.  Are you not using the front-facing camera on your phone?  Consider either putting a piece of tape over it or rigging a little flap for it on your phone case.  Either way, you want to avoid any recording cameras.  The more detailed the image and the more specific the parameters, the more precise is the recognizing ability.  A high-resolution camera at an ATM or the main entrance of a store, for instance, will capture a better image than a camera scanning a crowd on a city street.  When using an ATM or checking out at a cash register at a grocery or retail store, look down, so the brim of your hat covers the points around your eyes.  Consider a cloth mask or scarf with a high contrast pattern on it.  Consider clear acetate tapes or mylar sewn into your clothes and hat to decrease how clearly cameras see you.


Eluding the capture of your clear image by facial detection and recognition software can be done with a conscious effort and a few tricks.  If you imagine the old Hollywood movies where the hero tries to hide with dark glasses, a scarf, and a hat, some of this isn’t too dissimilar; however, patterns, materials, and creative ways to conceal your facial markers will keep your unique face out of databases.

What do you think?  How important is it that we start protecting our faces from invasive facial detection and recognition programs?  Is it the biggest threat we face? 

As always, please stay safe out there.

5 4 votes
Article Rating
Notify of
Inline Feedbacks
View all comments

Related Posts

Would love your thoughts, please comment.x