Xilinx has developed two new Edge AI demos for tradeshow and customer use. These represent good real-world use cases of Xilinx AI technology in action. The first is an 8-channel emebedded Vision + AI demo. The second is a demo containing six simultaneous Nerual Nets in an Automotive use case. 

8-Channel Embedded Vision + AI Demo

This demonstation design features an 8-channel VCU + AI in single Xilinx Zynq UltraScale+ Embedded Vision device. It performs both face and traffic analytics using the Xilinx DPU (AI hardware overlay) with eight different RTSP video streams. It’s a great real-world example of AI-based edge computing enabled by Xilinx!

Six Simultaneous Neural Net Model Demo

This demonstration features six different CNN models running simultaneously on single Xilinx Zynq UltraScale+ device. Two MIPI cameras plus two video files are used as the input to the DPU). It demonstrates good performance and low end-to-end latency (<200ms) for a real-world, multi-network application. It’s another great illustration of how Xilinx solutions deal with time-sensitive visual perception in a challenging environment such as autonomous driving.

For more information about these two demos, please contact your Xilinx sales representative.