Uncanny AI Box for ALPR Application with Kria KV260 Starter Kit
With the public announcement of the Kria SOM portfolio in April, came many new edge applications on the Xilinx App Store. As a one-stop-shop platform that hosts, markets, and sells great applications from Xilinx ecosystem partners, the App Store now offers accelerated apps & IP cores for both edge and cloud. Uncanny Vision is among one of the earliest Kria SOM users that developed their own ALPR solution with the KV260 starter kit and is now selling on the Xilinx App Store.
Uncanny ANPR is used to develop an end-to-end AI-based license plate recognition system that delivers high accuracy even in challenging conditions such as fast-moving vehicles, different fonts, dirty plates, difficult camera angles, and so on. With the Kria KV260 starter kit, they were able to achieve a deep learning-based AI box for multiple real-time camera streams that detect, track vehicles, and reads license plate text. Powered by Vitis AI 1.3, the system takes full advantage of Xilinx DPU acceleration and deploys multiple AI models for the detection of vehicles, license plates, vehicle classifications, and license plate recognition. In addition, it is featured to process up to 2 streams of H.264/H.265 decode at 1080p resolutions. The end-to-end system is also designed to provide dashboards, remote monitoring, blacklist and whitelist of vehicles, open REST APIs for custom integration, and so on.
If you are looking for a flexible and easy solution for your applications that include smart parking, gate security, traffic management, or similar scenarios, check out the Uncanny AI Box application on the Xilinx App Store and our network of ecosystem partners will be connected with you to provide you the best technology decisions with your specific needs.
