Taking traffic monitoring forward

The rise of smart traffic applications and video surveillance systems have led to a need for automatic number plate recognition that is highly accurate. It is a requirement that Avnet’s automatic number plate recognition system can more than easily meet.

Developed by Avnet Engineering team, this license plate recognition design is based on Avnet Ultra96 board, a cost-optimized Xilinx 16nm SoC prototyping platform for embedded vision and industrial IOT systems. It is a reference design that deploys the custom deep learning network in Tensorflow framework together with Xilinx AI development tools and IP, achieving the result of accuracy performance of 92% and a recognition performance of up to 37.6 plates/second.

Car driving at night with LED headlights


  • Custom Deep Learning Network
  • Tensorflow framework
  • MMdnn framework conversion
  • Xilinx DeePhi DNNDK 2.08
  • Xilinx DeePhi DPU
  • Accuracy 92%
  • Performance 37.6 plate/s

Target applications

  • Parking automation
  • Electronic road tolling

Key components

  • Avnet Ultra96 development board
  • Xilinx Zynq UltraScale+ MPSoC ZU3EG
  • 2GB LPDDR4
  • Wi-Fi
  • USB Camera


Guide Automotive solutions - Grid Light

Display portlet menu

Reference Solutions

Automotive solutions

From connected cars, security, safety to energy saving, our expertise in automobile will support your innovations, enable interactions between cars, people and infrastructures.

Contact Asia Design Services

If you are interested in this solution, please contact us. Stay up to date with our latest news on Reference Solutions.

Contact Us