Deploying model on raspberry pi using yolov8

Hi, I’m trying to deploy a custom obj detection model using yolov8 on a raspberry pi 4 bullseye 64 GB. I have PyTorch installed of course, and used the docker workaround as explained in the docs to deploy on a raspberry pi.

I am able to download the dataset using the code snippet for yolov8, but when I run the yolo command using my selected .yaml data file it freezes up and the job is eventually killed when it begins training for n number of epochs. It isn’t letting me attach multiple images here but I will put an imgur link will all the relevant screenshots.


What stuck out to me was the line saying “using 0 dataloader workers” and I’d presume this is the root of the issue, but I of course aren’t sure. My thinking is to maybe install some special PyTorch wheel, but i’m unsure if any exist and am just at a wall here. Any help would be greatly appreciated!