In this article we showcase the DUO platform running across several ARM powered devices. This includes a Nvidia Jetson TK-1 Devkit, a Radxa Rock dev board and the upcoming compact DUO Vision PC device. Here is a video showcasing the ARM support and our latest software updates.
The DUO Dense3D middleware provides a simple C/C++ API for extraction of high performance dense disparity and depth information from stereo frames and saving texture and depth data into Polygon File Format (PLY) file. The middleware is optimized to run in real-time and is paired with the Dense3D app which loads into the DUO Dashboard for quick preview and testing. Moving onto the hardware, we start with the NVIDIA board:
NVIDIA Tegra K1 – Jeston Board
We start with the latest developer board from Nvidia (Jetson), offering a powerful Tegra processor showcasing our middleware and Dashboard solution. The Nvidia Tegra offers an amazing amount of processing power in a compact package and we can’t wait to see what exploring it with our customers and partners. Strengths: Quad Core CPU with 192 cores, CUDA, Linux, DDR3 RAM. We had a real blast working with this board as it performance-wise feels and acts like a real desktop computer.
Radxa Rock – Devboard
Next we showcase the Radxa Rock dev board which features a quad core RK3188 ARM chip. We have chosen this board for its great documentation, open source design and the ease of use. Strengths: Quad core ARM, Mali GPU, DDR3 RAM, bootable from uSD card. We loved working with this board because of the good support and well as high level of hack-ability. Performance-wise it is slower than the Jetson board but on the other hand it consumes much less power than Jetson.
DUO Vision PC
And finally we preview the DUO Vision PC which ships with a custom built and highly tuned distribution of Linux kernel perfect for computer vision research and beyond. This allows for a full computer vision solution that fits in the palm of your hand and comes pre-bundled with algorithms and tooling for developing your applications directly on the device. Strengths: Custom optimized DUO Linux OS, Quad Core ARM, Mali GPU, DDR3 RAM, bootable from uSD card. We are extremely excited about this device and can’t wait to see what you will do with it.
We believe that ultra compact mobile vision systems are a requirement for enabling and discovering the next generations of computer vision and 3D perception. And are really excited about the next generation low power processors as well and will be sharing updates as we test new devices. Our goal is to make an easy to use and powerful vision platform that runs anywhere. We are realizing this with the DUO Development Kit (DDK1) and continues updates to the platform.