Instead of going as Edward Snowden or digging out that tattered skeleton outfit, you could simply apply a little makeup to become a new kind of Halloween ghost: The Person Impervious to Face-Recognizing Cameras.
According to a helpful makeup video on YouTube, all you need is some tape, scissors, eyeshadow and something called CV Dazzle. Developed by the artist Adam Harvey, CV Dazzle is a modern take off on the dazzle camoflage from World War I, which was used to protect warships from submarines. This one is designed to hide from cameras ("CV" means computer vision). The video is a take off too, by an artist named Jillian Mayer, and is just as deadpan. "The more you distort your look" by applying the bright paint around the nose, bridge, lips or ocular region around the eyes, she says, "the harder time computers will have identifying you." Face paint works too. (Apparently the camoflage of the future will be wildly visible).
A couple of years ago, Adam Harvey and DIS Magazine showed off some of the possibilities:
Photos courtesy DIS Magazine
Computers are slowly getting better at detecting (unobscured) faces, and the companies that compile the most photos of us are keenly interested in face recognition technologies. Facebook, which bought Face.com in 2011, was chided by Congressmen into blocking the feature until just a few months ago, when it started encouraging face tagging; Google has rolled it out slowly too, but has said it will prohibit face rec tech from working on its computer glasses (though there's no certainty a ban would work).
Facebook limits its face recognition tagging system to one's friend network—you aren't allowed to identify people who aren't your friends—and plus, the math may be just be too big for Facebook-wide face searches, as Neeraj Kumar, a face recognition researcher, recently told NPR. But if one could cross check faces across Facebook's database, that would make, he said, for "a useful search tool."
“I think most people truly expect to be anonymous in public,” Jennifer Lynch, an attorney for the Electronic Frontier Foundation, told Motherboard recently. Anonymity is a crucial element of public spaces and communication, she argues, and allows us to freely participate in a democratic society. “[When people] are being watched,” she said, “it chills their behavior—it chills who they will associate with, how they will act, and what they will say. That’s not a society I want to live in or raise my son in.”
The software is already there, waiting for better inputs (read: legible faces). As my colleague Brian Merchant reported in August, a Homeland Security system for face recognition in crowds could be "five years" from active service (it's called BOSS). The FBI is working on its own face reading system, one that could allow law enforcement agencies to identify subjects in “public datasets" like Facebook. (The FBI has been sued over the program's transparency; currently, there is legislation to limit the use of biometric technology in three states, Illinois, Texas, and Washington.) Already, the US claims one of the largest programs—the State Dept. operates a database of about 75 million photos to cross-check visa applications—and face recognition systems are in use by governments around the world.
"The technology is currently at a state where these face recognition algorithms can be deployed in anything," one facial recognition expert told NPR, "from cell phones to large multiserver search engines capable of searching over 100 million faces in just a few seconds with operational accuracy."
That is, of course, so long as you're not wearing your makeup. And don't forget your infrared glasses and drone-proof cloak.
More on algorithmic recognition: