Science

This Halloween, Dodge Surveillance with Face Camouflage

Giphy/YouTube/Fusion

If you want a costume that screams the future of governmental control is surveillance, but think Elliot from Mr. Robot is too obvious a choice, then consider painting stripes on your cheeks, and blocking out your nose. Sure, some might confuse you as a resident of the Hunger Games’s Capitol, but joke’s on them — what you’ll really be, is a master at bypassing facial recognition software.

With his ‘CV dazzle’ methodology, named after a type of World War I naval camouflage artist, Adam Harvey uses careful placement of makeup to break up the facial recognition cues surveillance softwares rely on. Harvey describes this as creating an “anti-face” — a disruption of the features, symmetry, and contours of a human mug. Just as the human brain is confused by a fake beard, surveillance software can be thrown off by some choice makeup decisions.

Harvey designed CV dazzle to outwit the Viola/Jones method, a real-time object detection system that analyzes faces by calculating its different spatial regions. Most facial recognition systems look at “key landmarks,” like the nose bridge, eye corners, and chin, which are used to create a depth map. By making your face look asymmetrical, covering your nose bridge and ocular regions, and obscuring the elliptical shape of your head, you too can avoid detection.

The CV dazzle technique is just as good at dodging surveillance as a mask. It might even be better, considering it lets you dodge the anti-mask laws of certain states. When this camouflage makeup technique was applied to the faces of celebrities, and processed through eight different facial recognition systems, a study from the University of Texas at Dallas found, 83 to 100 percent of the time, the system could not recognize the celebrity.

Regardless of your reasons for going as a “surveillance hacker,” it’s likely that on the night of Halloween you may really be obscuring the surveillance systems around you. After all, the FBI’s “Next Generation Identification” system has access to approximately 411 million face photos — and there’s only 319 million people in the United States. As visual performance artist Jillian Mayer explains in her tutorial, with this look, “you’ll be undetectable by cameras, and you’ll look great.”