基于PYNQ-Z2人工智能开发平台的人脸追踪检测系统设计

文章提出一种基于赛灵思嵌入式人工智能FPGA开发平台PYNQ-Z2的人脸追踪检测原型系统。通过搭建Linux环境并运行Jupyter Notebook交互式在线编译器,调用IPython内核和PYNQ的硬件库,导入OpenCV库编写并且编译执行Python程序,可实现在线控制PYNQ-Z2来获得视频流输入并对视频数据进行传输和处理,从而实现人脸图像的分析与检测。本设计利用了PYNQ-Z2的可编程系统(PS)和可编程逻辑(PL)的交互式数据传输处理,尽可能的发挥ARM和FPGA的优势,从而实时的进行人脸检测。

This paper presents a prototype system of face tracking and detection based on PYNQ-Z2, an embedded artificial intelligence FPGA development platform. By building Linux environment and running Jupyter Notebook interactive online compiler, invoking IPython kernel and hardware library of PYNQ, importing OpenCV library, compiling and executing Python program, we can realize on-line control of PYNQ-Z2 to obtain video stream input and transmit and process video data, so as to realize face image analysis and detection. This design makes use of the interactive data transmission processing of PYNQ-Z2’s programmable system(PS)and programmable logic(PL), giving full play to the advantages of ARM and FPGA as far as possible, so as to carry out real-time face detection.