I guess I'm easily distracted. I've been struggling a bit trying to formulate a plan to accomplish my project for Path II Programmable and in my searching I saw that a new V2.5 Pynq image for the Ultra96v2 was just released: . Since I've used Pynq on the Pynq-Z2: PYNQ-Z2 Dev Board: Python Productivity for Zynq
- Review , I decided that it might be a nice diversion to try it out on the Ultra96v2.
I downloaded the image and loaded it on a new micro SD card and booted up. I think there's something about the Ultra96 WiFi AP implementation that doesn't like me because I couldn't connect again but this time it didn't even look like an AP was configured (I connected through the JTAG/UART port). I was going to use the upstream USB3 port to connect to the Jupyter notebook via Ethernet over USB but remembered that I had a TrendNet USB3 to Ethernet dongle that I had not tried out.
The dongle worked great and connected immediately. I was able to log into the Jupyter notebook and used the wifi.ipynb notebook to also connect to my WiFi router.
Adam Taylor had done a post on Microzed Chronicles about Pynq on the Ultra96 https://www.hackster.io/news/microzed-chronicles-ultra96-and-pynq-da3b22cc982 and I followed along to get the Pynq Computer Vision examples from the Xilinx Github repository https://github.com/Xilinx/PYNQ-ComputerVision . You can see it loaded in the Home directory below (pynqOpenCv).
I decided just to run through the Intro examples for now. All the examples require a webcam as an input video source so I attached a Logitech C525 to the remaining USB3 port on the Ultra96.
Here's the frame capture:
After running filter2D:
After running dilate:
Running both filter2D and dilate and profiling:
Profiled results:
For these examples the OpenCv functions are running in SW. Being able to run interactively should help me figure out how to implement some of the OpenCV functions that I'll need for my PIIP project. That's my hope anyway......
