Immediate System Response Speed
reVISION for Rapid Deployment of Algorithms and Sensors

Xilinx participated in the Automotive Innovation Day 2017 held on the 5th and introduced Xilinx's automotive solutions.

Xilinx showcased reVISION, which features immediate system response speed and is suited for markets requiring rapid deployment of algorithms and sensors. The reVISION stack facilitates the development of intelligent vision-based systems including machine learning, computer vision, and sensor fusion. It supports the most commonly used neural networks such as AlexNet, GoogLeNet, SqueezeNet, SSD, and FCN.




For application-level development, Xilinx enables the use of Caffe for machine learning and OpenVX for computer vision. Additionally, it provides library elements for CNN network layers necessary to build custom neural networks (DNN/CNN).

At the venue, reVISION was demonstrated. Using a stereo camera with HDMI output to represent depth through color gradients, showing motion speed and direction, and incorporating deep learning technology to distinguish between people and objects—all stacked into one solution. Furthermore, in collaboration with LogicBricks, Xilinx showcased an AVM system that seamlessly stitches together video feeds from four cameras.