Fast, portable, power-efficient vision processing across diverse hardware architectures; Full conformance test suite and Adopters Program immediately available; Khronos to ship open source OpenVX implementation by end of 2014
BEAVERTON, Ore. — (BUSINESS WIRE) — October 20, 2014 — The Khronosâ„˘ Group today announced the ratification and public release of the finalized OpenVXâ„˘ 1.0 specification, an open, royalty-free standard for cross platform acceleration of computer vision applications. OpenVX enables performance and power-optimized computer vision processing, especially important in embedded and real-time uses cases such as face, body and gesture tracking, smart video surveillance, advanced driver assistance systems (ADAS), object and scene reconstruction, augmented reality, visual inspection, robotics and more. In addition to the OpenVX specification, Khronos has developed a full set of conformance tests and an Adopters Program, that enables implementers to test their implementations and use the OpenVX trademark if conformant. Khronos plans to ship an open source, fully-conformant CPU-based implementation of OpenVX 1.0 before the end of 2014. The full OpenVX 1.0 specification and details about the OpenVX Adopters Program are available at www.khronos.org/openvx.
OpenVX defines a higher level of abstraction for execution and memory models than compute frameworks such as OpenCLâ„˘, enabling significant implementation innovation and efficient execution on a wide range of architectures while maintaining a consistent vision acceleration API for application portability. An OpenVX developer expresses a connected graph of vision nodes that an implementer can execute and optimize through a wide variety of techniques such as: acceleration on CPUs, GPUs, DSPs or dedicated hardware, compiler optimizations, node coalescing, and tiled execution to keep sections of processed images in local memories. This architectural agility enables OpenVX applications on a diversity of systems optimized for different levels of power and performance, including very battery-sensitive, vision-enabled, wearable displays.
â€śIncreasingly powerful and efficient processors and image sensors are enabling engineers to incorporate visual intelligence into a wide range of systems and applications,â€ť said Jeff Bier, founder of the Embedded Vision Alliance. â€śA key challenge for engineers is efficiently mapping complex algorithms onto the processor best suited to the application. OpenVX is an important step towards easing this challenge.â€ť
The precisely defined specification and conformance tests for OpenVX make it ideal for deployment in production systems, where cross-vendor consistency and reliability are essential. OpenVX is complementary to the popular OpenCV open source vision library that is also used for application prototyping but is not so tightly defined and lacks OpenVX graph optimizations. Khronos has defined the VXUâ„˘ utility library to enable developers to call individual OpenVX nodes as standalone functions for efficient code migration from traditional vision libraries such as OpenCV. Finally, as any Khronos specification, OpenVX is extensible to enable nodes to be defined and deployed to meet customer needs, ahead of being integrated into the core specification.
â€śAMD is an enthusiastic advocate for natural user interfaces enabled by computer vision,â€ť said Greg Stoner, senior director of application engineering, Heterogeneous Applications and Solutions, AMD. â€śAs a proponent of open standards and a provider of highly-parallelized architectures ideal for computer vision, we embrace the OpenVX standard and look forward to standardized proliferation of these experiences throughout the industry.â€ť
â€śOpenVX will be a great tool for future development of computer vision applications,â€ť said Johan Paulsson, CTO at Axis Communications. â€śWe see excellent potential for making development efforts future-proof by separating algorithms from hardware.â€ť
â€śCadence is integrating OpenVX into our TensilicaÂ® Imaging/Vision Library software development kit to enable higher performance and power optimization for our scalable and configurable IVP-EP DSP cores, which are widely adopted for imaging, computer vision and automotive driversâ€™ assistance applications,â€ť stated Dr. Chris Rowen, CTO of the IP Group, Cadence.
â€śCEVA applauds the Khronos Group on achieving this important milestone with the release of OpenVX specification,â€ť said Eran Briman, vice president of marketing at CEVA. â€śWith an extensive list of vendors currently sampling their computer vision ICs based on the CEVA-MM3101 DSP and a growing developers community, it is clear that the optimal way to meet the stringent performance and power requirements of computer vision applications is to offload CV processing to a dedicated processor. OpenVX enables this for developers in a seamless manner and we are now integrating it into our CEVA-MM3101 Application Developers Kit (ADK).â€ť
â€śAs one of the first innovators in embedded vision, CogniVue will be supporting a compliant OpenVX implementation for our APEX Image Cognition Processing technology with announcements coming,â€ť said Tom Wilson vice president of product management and marketing at CogniVue.
â€śWe see OpenVX as a promising tailor-made standard tool for developing
performant, low-power, while portable computer vision applications for
our mobile devices,â€ť said Zhouhong, president of Central Hardware
Engineering Institute, Huawei. â€śBased on its tightly defined
nature, we also see OpenVX to serve as a standard benchmark suite which
can promote the development of ever-more power efficient computer vision
accelerators, a component likely to become must-have for coming mobile