Cambridge Invents // Kinect & computer vision - Collusion

Cambridge Invents // Kinect & computer vision

Collusion’s 2014 series of films celebrating Cambridge’s creativity

Press play above to hear from Jamie Shotton about his role in creating the Kinect and the challenge of computer vision.

Do you know that Microsoft Research, Cambridge, played a vital role in the development of Kinect for Xbox, or that Kinect held the Guinness World Record in 2011 as the fastest selling consumer electronic device in history?

Kinect launched in 2010, selling 8 million devices in its first 60 days on the market. Most commonly you find it with an Xbox, allowing users to play games without the use of a controller, instead via gesture and voice commands. One of the hardest problems in computer vision has been to detect the complexity of human movements, based on physical size, shape and distance of the body from a camera sensor. Microsoft solved many of these problems using an infra-red projector and camera to track movement, using them to teach a computer to learn whose body parts belong to whom.

During its development, the research team working on Microsoft’s top secret ‘Project Natal’ (code name for Kinect) carried out thousands of observations of human movements in homes and even in Hollywood Studios, to create a complex set of movements to teach and test the early iterations of the Kinect system.

Project Natal brought together a diverse team of talented individuals from across Microsoft, including Jamie Shotton. By 2014, a Senior Researcher at Microsoft Research, Jamie’s PhD at the University of Cambridge working on Computer Vision and Visual Object Recognition proved to be a key component in the development of Kinect. In 2008 he received an email from Microsoft colleagues in the United States asking him to help them with Project Natal. Collaborating with other colleagues at Microsoft Research in Cambridge, this ask resulted in direct work that produced a machine learning algorithm for human motion capture in Kinect.

Microsoft has since opened up development for Kinect by the release of Kinect for Windows, allowing people to invent their own uses for the device. As a result, people have used Kinect in operating theatres, to play virtual pianos, create 3D maps of rooms and someone is even creating a user controlled interface like the one in the 2002 film, Minority Report. Often science fiction falls flat in predicting the future, but in this case gestural interfaces, the ability to control computers through human motion, are now common in the domestic space. Whilst they’re still in their infancy, it’s exciting to discover that such powerful breakthroughs were contributed to by a talented team at Microsoft Research in Cambridge.

Press play above to hear from Jamie Shotton about his role in creating the Kinect and the challenge of computer vision.