Social Disatancing Monitor using yolov3 and DPU HW acceleration for Xilinx adaptive computing challenge 2020
Hardware layout：
Image test for functionality：
You should follow the steps below to reproduce it.
- Download PYNQ image v2.5 from https://github.com/Avnet/Ultra96-PYNQ/releases .
- Image the SD card and boot the device, with USB camera attached, USB to Ethernet adapter to connect to Internet and a mini-DP to DP cable connecting to a Monitor.
- Follow the instructions on DPU-PYNQ page to set up the examples(Xilinx/DPU-PYNQ: DPU on PYNQ (github.com)) by getting all the notebooks, go to this page(DPU-PYNQ/README.md at master · Xilinx/DPU-PYNQ (github.com)) , and download the .hwh file for Ultra96v2 to employ DPU acceleration.
- Run the dpu_yolo_v3.ipynb example project and make sure there is no warnings or errors when executing this ipynb file.
- Shift to this project, run one by one and you will get the right results.
If there is one python package missing, please install it on your own. (for I almost forget when I install the other used packages that are not installed when installing dpu-pynq)
- improve performance from Hardward(utilize more hardware resources like resize and more powerful DPU) and Software(reading USB camera is I/O block job).