CEVA, Inc. has announced CEVA-CV, a new computer vision (CV) software library for the development of vision-enabled applications targeting mobile, home, PC and automotive applications. Optimized for the CEVA-MM3101 imaging & vision platform, the software library enables application developers to simply and efficiently add vision capabilities to System-on-Chips (SoCs) incorporating the CEVA-MM3101.
CEVA-CV is based on OpenCV, a standard library of programming functions for computer vision processing. Including more than 500 of the most frequently used computer vision functions, CEVA-CV contains key CV kernels and algorithms required for application developers targeting mobile computing, PCs, smart TVs, natural user interface (NUI) devices and advanced driver assistance systems (ADAS). These functions include both simple CV kernels such as filters, histogram and affine transform as well as sophisticated CV algorithms including FAST, RANSAC, Connected Components and MSER. The complete set of libraries has been fully optimized using CEVA's C-level compiler for the CEVA-MM3101 imaging & vision platform. This significantly shortens development time and simplifies the development process by eliminating the need for the programmer to understand the processor architecture, allowing them to develop in high-level languages.
"The introduction of the CEVA-CV computer vision libraries is the result of more than 100 man years of engineering invested in developing computer vision platforms and applications," said Eran Briman, vice president of marketing at CEVA. "Together with our partners and customers in this rapidly developing technology field, we identified more than 500 of the most critical functions to support, and adapted these to ensure maximum power efficiency in embedded applications. Computer vision software programmers can easily add these functions to any CEVA-MM3101 enabled SoC and leverage the CEVA-MM3101 vector processor engine and C-level compiler to enable outstanding performance for their target application."
"The addition of CEVA's OpenCV-based computer vision libraries to the CEVA-MM3101 imaging and vision platform is a significant step forward for the development of embedded vision applications," said Jeff Bier, founder of the Embedded Vision Alliance (http://www.Embedded-Vision.com). "Embedded vision applications are diverse and performance-hungry. By providing a large library of optimized software building blocks, CEVA is enabling application developers to more easily create embedded vision applications on SoCs based on the CEVA-MM3101. I commend CEVA on this exciting and important development in their computer vision roadmap."
The CEVA-MM3101 is a unique, fully programmable platform dedicated to addressing the extreme computational needs of the most sophisticated image enhancement and computer vision applications. By off-loading the device's main CPU and replacing multiple hardwired accelerators for performance-intensive imaging and vision processing tasks, the highly-efficient CEVA-MM3101 dramatically reduces the power consumption of the overall system, while providing complete flexibility. Target applications for the CEVA-MM3101 include advanced image enhancement applications (including super resolution, HDR and video stabilizer), NUI applications (including gesture recognition, face recognition and gaze detection), ADAS applications (including forward collision warning- FCW, lane departure warning – LDW and pedestrian detection – PD), computational photography, and video analytics. For more information, visit www.ceva-dsp.com/CEVA-MM3101.html.
News Source: CEVA