Your online photo collection could soon be used to disguise strangers.
Scientists at Columbia University have come up with facial swapping software that uses publicly posted pictures as a kind of high-tech camouflage. The concept’s being suggested to Google for its Street View program. The team believes its creation could provide a better alternative to blurring faces on Google’s street shots — a practice started a couple of months ago to protect people’s privacy.
The program works by randomly selecting a face from a database of more than 33,000 photos, pulled from sites such as Flickr and Yahoo Images. It then creates a composite, taking the eyes and nose from the new image and placing them on top of the original face. The software is smart enough to match lighting, pose, and resolution so the parts don’t look out-of-place. It then aligns the features, color corrects, and blends the new face together — and the whole thing happens without any human guidance.
The results are a bit jarring at times, but they definitely do their job. The scientists are looking at offering the software to military or police groups for identity protecting purposes. It could also be marketed for everyday use, letting you switch out a bad grin with a more pleasing shot of your face — or, if you prefer, someone else’s.