there: Software from Prism Skylabs processes surveillance footage to
show how busy a place is without compromising customers’ privacy.|
"There's a lot of wonderful information locked up in video, and 40 million security cameras in the U.S. collecting it, but it's data that's not been available," says Steve Russell, cofounder and CEO of Prism, based in San Francisco. "We want to free up that information."
Prism's software can count people that come into a business, measure the length of the line at checkout, and produce static or animated visualizations showing how people moved around a store. It is designed so that it cannot identify or track individuals. One national wireless carrier is already using Prism's technology to generate heat maps of where visitors go in their showrooms, to compare the level of interest in different devices—valuable data to them and to the device makers.
Prism's software can also be used to turn security footage into a live version of Google's Street View, says Ron Palmeri, Prism's president and other cofounder. "We give the ability to go beyond the facades of businesses and show you the inside and even how busy it is, using very effectively privacy-protected imagery."
That imagery can show people blurred into anonymous ghosts, in what Russell calls "Predator vision" (a pixelated image), or have people disappear altogether to be replaced with a "heat map," on which colors signal the density of people. One gym in San Francisco trialing the technology plans to use it to show customers a live view of how busy it is.
Although security cameras are typically low quality, Prism uses computational photography techniques to combine multiple frames to produce images with higher quality and resolution than the original video. "We can remove the ugly, grainy quality of surveillance footage," says Russell.
Prism's software is designed to be used with existing security cameras. Software installed on a computer linked to the cameras digests the raw video into a compressed form that is sent to cloud servers, where Prism's software does the hard work. It sends back the visualizations, statistics, and other data it extracts to a PC, smart phone, or tablet.
Since surveillance cameras are static, Prism's software can work out which parts of a scene are fixed in place, such as walls, flooring, and furniture. Anything that moves against that background is clipped out and subjected to further processing. Russell hints that he hopes to eventually offer more sophisticated analysis by taking advantage of emerging computational photography techniques—for example, Lytro's light field cameras, which produce images that can be refocused after they have been taken.