Venture Firm Realizes High - Speed, Low - Delay Gesture - Based Ui



Venture Firm Realizes High-speed, Low-delay Gesture-based UIExvision Inc, a venture firm spun off from the University of Tokyo, developed a technology to realize a high-speed gesture-based controller by using a conventional CMOS sensor.



The technology, which the company calls "Exvision Gesture System (EGS)," can recognize both hands of the user at a frame rate of 100fps. It is targeted at smart TVs, game consoles, intellectual training toys, digital signage, medical devices, etc.

The new technology uses a conventional CMOS sensor. Therefore, it can realize a gesture-based controller at a lower cost, compared with the Kinect, which uses a range image sensor, and stereo vision devices. Also, because of the high frame rate, it realizes a user interface (UI) that enables smooth operations and does not often lose track, Exvision President Akio Fujii said. Its delay time is about 30ms though it depends on conditions.

The EGS supports two-point multi-touch UIs and enables to control Windows 8.1- and Android-based devices, etc by making gestures with both hands. A clicking (tapping) operation can be performed by making a fist. And it supports other gestures, too.

The EGS will be provided in two forms: (1) a software system for personal computers and (2) a camera module. Because the EGS uses a GPU for the recognition of hands, the first option is available to personal computers, etc that are equipped with a GPU. And the second option is for personal computers, etc that are not equipped with a GPU. The camera module comes with an SoC capable of image recognition and supports USB connection.

The license price of the software system and the price of the camera module have not been decided yet. But samples of the module are scheduled to be shipped in the fall of 2014.

Exvision is a venture firm financed by Innovation Network Corporation of Japan (INCJ), a Japanese government fund. It uses the research results of Ishikawa Watanabe Laboratory at the University of Tokyo. And the image recognition algorithm of the EGS is part of the results.

Photo Gallery

Venture Firm Realizes High-speed, Low-delay Gesture-based UI Venture Firm Realizes High-speed, Low-delay Gesture-based UI