Press release: new low-power vision and video processing solution

Last week, we announced our new unified computer vision and video coding solution. We’ve combined our low-power, multicore v-MP4000HDX architecture with our accelerated OpenCV library and a suite of Full HD and 4K video encoders and decoders. The processor is 100 times more powerful and consumes 1000 times less energy than host CPUs on video and vision applications, allowing for always-on camera applications, more accurate algorithms and an overall higher-quality user experience. The EETimes, EE Journal,Design&Reuse, Chipestimate and Embedded.com ran articles about our new solution.

The v-MP4000HDX processor family scales from 3 to 10 cores and can be licensed for inclusion in SOCs. This enables a slew of new applications for mobile, gaming and TV platforms: augmented reality, gesture interfacing, depth mapping and computational photography enable new user experiences and usage models. In automotive applications, the company’s low-delay, high dynamic range video codecs combined with computer vision techniques enable both camera-based advanced driver assistance systems and surround-view cameras to prevent collisions. The v-MP4000HDX architecture can simultaneously decompress, analyze and compress video data, enabling a wide variety of imaging and video applications.

Read the full press release.

Feature highlight – videantis OpenCV acceleration

Begun at Intel back in 1998, the OpenCV library has become one of the most popular tools in the computer vision industry, reaching 6 million downloads in early 2013. The library has more than 2500 optimized algorithms and is being used by companies such as Google, Microsoft, Intel, Honda and Toyota. OpenCV is free for use under the open-source BSD license, and is cross-platform, running under both PC-based and mobile operating systems and hardware, making it easy to move from a PC-based development system to an embedded consumer application. OpenCV is a great starting point for developing any computer vision application thanks to the wealth of algorithms included, as well as the scores of example applications available online.

Our OpenCV implementation typically runs on a host CPU such as an ARM Cortex-A series processor. With a few lines of code, the software developer can select those sections of the algorithm that she wants to accelerate on our subsystem. We take care of processing the image data on our v-MP4000HDX subsystem and accelerating the algorithms on the multicore processor. In contrast to GPUs, the videantis subsystem accesses the same memory as the host processor, removing the need for inefficient extra memory copy operations. Typical speedups are two orders of magnitude, and power consumption savings three orders of magnitude. And the software developer can implement all this just by changing a few lines of code.

Read more about our OpenCV implementation.

Vision tech of the month: Smart headlights shine between the raindrops

In recent years, Carnegie Mellon has been working on a very impressive computer-vision-enabled automotive application: smart headlights that shine around raindrops or snowflakes. The net effect is to greatly improve visibility: raindrops and snowflakes no longer reflect the light from your own headlights back into your eyes, making your view of the road much better.

In June, Edmunds highlighted the system as one of the six new car technologies that will simplify our lives. One thing’s for sure: to put such vision-based automotive safety systems into production, you need a flexible, high-performance vision processing subsystem such as our v-MP4280HDX processor (pdf).

We’re hiring!

The videantis team is growing. We’re looking for a field applications engineer with a passion for video, as well as a sales manager who can work large semiconductor accounts on a global scale. Are you interested in working with top-notch people and top-notch technology in a rapidly growing market? We’d love to hear from you.

View our open positions