Remember Dazzle, the method of complex geometric camouflage painting introduced in WWI (designed to disguise the direction, size, and types of ships at sea)? Last year, ITP student Adam Harvey’s Dazzle CV project brought a similar technology to modern facial recognition software. Using hair and make-up tricks, Adam realized you could fairly easily defeat facial recognition algorithms found in CV software and face-recog bots like the ones that Facebook and Flickr uses. And, you get to look like a neo-tribal cyberpunk in the process. He describes CV Dazzle as “antagonistic technology.” Eye of HAL, your move. [Thanks, Jake!]
10 thoughts on “Computer Vision Camo: CV Dazzle”
Comments are closed.
Brilliant– as Keith just pointed out (offline), we’re willingly building the database of faces ourselves. We wondered if some day pupil recognition could be done at a distance, even on someone who’s wearing a mask. Goli just saw people in the Amsterdam airport do express check-in by looking into a pupil-recognition camera.
It is ironic that the way to make yourself unrecognizable to computers makes you unforgettable to humans. I wonder if just sticking on a couple of stickers with eyes on them near your real eyes would be effective.
Thanks for sharing such an interesting article to the readers. I just love this and also Have bookmarked this blog. Love to see more from you. Keep it up. Best Wishes. :)
PC Support
Hello Everyone, Looks like some of you might be interested in this opportunity.
We are looking to fill in a part time/full time telecommute “COMPUTER VISION EXPERT” position. For more details/apply click this link –
http://hire.jobvite.com/j/?cj=o3o6Vfwc&s=Kamla_Web
Soooooo, the algorithm should be modified to detect face camo possibility leading to one of those TSA pull-asides at least at airports.