CEVA and eyeSight announce availability of software-based gesture recognition solution for the CEVA-MM3101 imaging and vision platform
News Release from:
28 June 2012
CEVA, Inc. and eyeSight Mobile Technologies today announced the availability of an optimized software-based gesture recognition solution for the CEVA-MM3101 imaging and vision platform.
The combined solution is targeted at a broad range of smart devices including smartphones, tablets smart TVs, automotive infotainment systems and more. eyeSight’s software leverages the CEVA-MM3101 to offer substantial competitive advantages over alternative implementations of gesture recognition-based technology. As an example, the combined CEVA-eyeSight solution requires 20X less power consumption compared to an ARM Cortex-A9 based alternative solution.
Jeff Bier, founder of the Embedded Vision Alliance commented: “I commend CEVA and eyeSight for their leadership in advancing the deployment of embedded vision technologies in cost- and energy-sensitive applications. Their combined solution brings together optimized hardware and software technologies to deliver a platform for vision-based gesture recognition in any camera-enabled device. Licensable solutions such as this are crucial for the adoption of embedded vision technologies in mass market devices.
Vision-based gesture recognition is fast becoming one of the most sought after technologies in the mobile and digital home industries, offering both semiconductors and OEMs a new area in which to differentiate their products and enable new and value-added features. eyeSight’s human machine interface (HMI) is currently the most innovative interface technology available for embedded systems, utilising the existing camera in any smart device to enable touch-free control. The CEVA-eyeSight solution supports multiple hand gestures as well as palm detection and tracking, supporting multiple users, up to 4.5 meters in distance away from the device and in low light conditions. The entire solution, incorporating eyeSight’s software and the CEVA-MM3101, requires less than 70MCycles for the complete gesture application, while consuming less than 20mW power in a 28nm process. To enable customers to further differentiate their solution, the fully programmable CEVA-MM3101 engine can be utilized to perform an array of additional vision-based functions in software, including face detection, eye tracking, 3D map creation, object tracking and image enhancement applications.
Gideon Shmuel, CEO of eyeSight, commented: “Our strategic partnership with CEVA has allowed us to further develop and refine our industry-leading gesture technologies by leveraging their powerful CEVA-MM3101 imaging and vision platform. Our combined solution offers the industry’s lowest power consumption for gesture recognition making it ideal for mobile devices. Furthermore, the ease with which we ported and optimized our gesture recognition software to the CEVA-MM3101 platform is a testament to the quality of CEVA’s commercial-grade tool chain and software development kit.”
Eran Briman, vice president of marketing at CEVA, commented: “Gesture recognition and other vision-based applications are the latest in a long list of consumer-driven applications that is fast becoming a must-have feature in smartphones, tablets and smart TVs. The processing required to perform these complex, real-time signal processing tasks is poised to place a significant burden on the performance and power consumption of battery-powered devices. The combination of the CEVA-MM3101 platform and eyeSight’s advanced software-based gesture recognition technology addresses these issues, enabling gesture applications far beyond the performance levels available today, while offloading the CPU from process-intensive signal processing tasks and significantly reducing the overall power consumption of the system. In addition, with more than 80% of the platform’s performance available for other applications, licensees can further differentiate their products to run additional vision-based as well as image enhancement functions.”