Winners Announcement: PYNQ-Z2: Embedded Vision Workshop (Zero to Hero) Series with Adam Taylor!

Not sure if you have seen it or not but there was a PYNQ project using YOLO on a PYNQ-Z1 board with Intel Movidius neural compute stick here:

Setting up the PYNQ-Z1 for the Intel Movidius Neural Compute Stick | FPGA Developer

That would be really nice for my number plate recognition project.

Because then I can finally figure out how to make my cat detection bot!

I have a project I've been working on using a FLIR Duo R IR camera on a SplashDrone 3 quadcopter with real-time IR video and sensor telemetry data.

The FLIR Duo R has HDMI out, and I'd like to add telemetry info embedded in the video stream.  I'm currently using some DJI CAN bus

on-screen-display modules, but I think I could do it better using the PYNQ-Z2 to capture the telemetry data and embed it onto the IR video stream.

The plan is to add some custom sensor arrays to the drone and add the sensor data to the video real-time.

Scott

The free WebPACK version of Vivado 2019.1 will work for the Zynq 7020 on the PYNQ-Z2.

it is weekend, i did some searching to see if others have done yolo with PYNQ Z2. In fact in in find some past road tester has done with tiny-Yolo (eg: Ralph's road test) , and other online material about yolov2 with custom hardware acceleration. In short, the higher performance yolo v2 probably can be achieved easier with some tinkering with custom overlay, and that's one attraction about this workshop, where it covers up to opencv in session. To be frank, there is a bit uncertainty, as opencv is by no way near or equivalent to yolov2 or its many siblings but at least i hope can understand more how to do custom overlay through the worskhop.

i am by no means an expert in yolo, so if you want to know more about the many yolo i mention above, do read about them online. here is one   , i am also just trying to understand them further.

https://medium.com/@amrokamal_47691/yolo-yolov2-and-yolov3-all-you-want-to-know-7e3e92dc4899

on my own proposed project, i have figured out that for ease of following the workshop, i will use my roku tv stick as hdmi input to the fpga board, then i mirror my roku to my android phone, then i will have an equivalent of  high resolution webcam to roam around.

i do not have such new version of vivado. just curious if we have the PYNQ-Z2 board, it is also comes with those voucher that tie our board to use Vivado 2019.1 as entitlement?

This is a really good online seminar that sums up my goals to learn Python, FPGA, and OpenCV for machine vision. I wish to develop a device that can convert written scientific data into digital text that can be stored into database systems. This would have application like in the link that follows:

https://arstechnica.com/science/2019/03/you-can-help-rescue-weather-data-from-the-1860s/

Old observations needs to be preserved as the paper records are deteriorating and remain unaccounted in scientific research because of accessibility.

I am really looking forward to join this workshop and thank you so much for arranging this.

thanks,

Sherdon

The reason is slightly different, different versions of PYNQ are tied to different versions of the design tools PYNQ 2v5 was developed using 2019.1 across the board as such we should develop using the same design tools. Otherwise slight IP changes etc might result in difficulties

Sounds like a great and generous workshop! I could see myself completing one of a few projects: one a synthesiser mapping and recognition tool to record synthesis performances. You provide a map of the controls and functionality in memory, and then using hand recognition, when a performer is interacting with the synthesiser, the board records what is being changed and when, mapping it out over time. Second idea is to build a synthesiser that outputs different sounds based on objects in the frame, or different patterns in the frame, so an instrument that is played by drawing!

thanks,

jack