Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

yolo v2 on raspberry pi 3

idata
Community Manager
624 Views

Hi expert

 

I am looking for a link to a github repository with version of yolo v2 that can run on raspberry pi 3.

 

I am runing ncsdk 2.05.00.02.

 

Thanks,

 

sojohans

0 Kudos
4 Replies
idata
Community Manager
300 Views

@sojohan You can use the ncappzoo version in the ncsdk2 branch of the appzoo, but you will have to do the Darkflow conversion on a PC in Ubuntu. Normally the appzoo tiny yolo v2 sample app clones the Darkflow github and does the conversion for you, but on the Pi Darkflow doesn't work. Here are the steps to do that:

 

     

  1. Clone the appzoo ncsdk2 branch to your PC running Ubuntu.
  2.  

  3. Go to the ncappzoo/tensorflow/tiny_yolo_v2 folder and do a make compile. You should see a graph file after this is complete.
  4.  

  5. Transfer the graph file to your Pi and put it in your ncappzoo's ncappzoo/tensorflow/tiny_yolo_v2 folder.
  6.  

  7. Run the tiny yolo v2 sample using python3 run.py
  8.  

idata
Community Manager
300 Views

Hi Tome

 

Super thanks

 

sojohan
idata
Community Manager
300 Views

Hi Tome

 

It worked. Do you know if there is a real time example?

 

sojohan

idata
Community Manager
300 Views

There isn't but you can just modify the code to use a camera instead of the image. That's about the only change you need to do. Please see the live image classifier sample and use the camera code from this sample for your project.

Reply