Many cities are packed with cameras pointlessly recording everything they see, but smart algorithms could allow them to keep only footage that matters
Lights, camera…oh never mind (Image: David Malan/Ocean/Corbis)
IT WAS being in London that got. The UK is awash with millions of closed-circuit TV cameras, and as Bahl walked around the city, he realised that much of what the cameras record would never be of interest to anyone.
“I thought, hey, there’s so much video streaming that no one’s looking at,” says Bahl, who is based at Microsoft Research in Redmond, Washington. Resources are being squandered all the time, on installingto record stuff that isn’t worth keeping. “We simply don’t have the ability to process all those images,” says Paul Schrater at the University of Minnesota in Minneapolis.
Bahl’s trip provided the inspiration for Vigil, an intelligent camera system to be presented next week at the International Conference on Mobile Computing and Networking in Paris.
Before anyone sees its footage, Vigil looks at every frame, counting the number of objects that might be noteworthy, such as people or car licence plates. Only then does it upload snippets of video, ranked from most to least important, to the cloud.
“It counts noteworthy objects like people or car licence plates, then ranks video clips accordingly”
For the last two months, the Vigil team has had the system running at three sites, surveying labs or office hallways in London, Redmond and Madison, Wisconsin. Next year they hope to launch a bigger pilot, perhaps monitoring traffic on a city road.
Bahl imagines that a system like Vigil could one day be used to detect the most exciting spots on the field during football games, switching TV coverage there automatically. It could also tell store managers when customers pick up certain products from their shelves.
This entry passed through the Full-Text RSS service – if this is your content and you’re reading it on someone else’s site, please read the FAQ at fivefilters.org/content-only/faq.php#publishers.