Getting started with Google Coral Edge USB Accelerator on Raspberry Pi 5: changing Python version

[@farhannaeem4186]

Thank you for putting this together! I had spent more than 2 days struggling to get my setup up and running and was starting to look for alternatives to Coral. I'm excited to finally have this up and running!

[@ExpatProfessor]

Well, I am glad it helped and now it runs!

[@Burgerhs]

I'm still struggling with mine I have till the 1st to return but I like it for my project but still can't do it

[@Levanterman]

Big thanks for taking the time to create this video , easy to follow process and sharing the info. My USB Coral is now working with the Pi5 🙂 😁👍

[@Cube_noob-cr7my]

When I run sudo apt-get install libedgetpu1-std I get this error:ModuleNotFoundError: No module named 'apt_pkg' Edit: Turns out that this isn't a problem. I did get an error about the numpy version when trying to run inference but 'pip3 install --upgrade numpy==1.19.3' fixed that. Now it's working great! Thank you for this video I'd been struggling for days.

[@abelrey1576]

Very good content, I'm glad to know that there are people uploading useful content like this. I look forward to seeing you implementing a yolo model with tflite and GCoral!

[@murraystaff568]

Fantastic video! thanks so much!

[@ExpatProfessor]

Glad it was helpful!

[@santiagoperez6769]

Great video my friend! I am enjoying your channel and learning tons... thank you! Question, why didn't you simply take the same approach as you did with the other video (from two months ago) where you used pyenv in your Ubuntu to manage the python versions? Is it not possible here, or maybe you just want to show multiple solutions? I am trying to follow along and I have a CM4 module so hopefully I have as much fun as you!

[@ExpatProfessor]

Well, that did not work. Still issues while using pyenv....

[@santiagoperez6769]

@@ExpatProfessor interesting... do you have any clues as to why pyenv isnt working in this case? I just flashed my CM4 module and installed the latest bookworm image, so I checked the python version and its the same as yours, i.e., 3.11.02, but I did not get a chance to try pyenv. I will try in the coming days and hopefully have something to report.

[@sandrocain-l2v]

can raspberry pi 5 and google coral usb run yolo models (like yolov8) well? I'm really looking forward to the next video if you are pleased.

[@ExpatProfessor]

Hi, YOLOv8 models are typically developed using frameworks like PyTorch or TensorFlow. To run these models on the Coral USB Accelerator, you would need to convert them to TensorFlow LITE format and then further compile them for the Edge TPU using the Edge TPU Compiler. So in other words, that is a lot of work.

[@sealmindset3840]

Does it matter what kernel version or is 6.1 good enough?

[@ExpatProfessor]

As of now, the current kernel version for RBPi is 6.1. I did not test it with any other. If you use the standard update and upgrade process, it will automatically update to this latest stable kernel version.

[@Burgerhs]

Hello Chief im new to this i wanna buy one Can you try a pre-trained roboflow model id like to see you? performance my surfaces pro 9 only get 6fps at 640x640 at 18fps at 100x100

[@ExpatProfessor]

Yes, Coral Edge TPU will accelerate a roboflow model. This is on my list of things to do!

[@Burgerhs]

@MachineLearningSebastian thank you so much for replying. I'll be looking forward for it. Thank you again

[@Burgerhs]

Lol, can you take control of my pc and install it for me? i desperately need it

[@ExpatProfessor]

This is Raspberry Pi tutorial 😀

[@Burgerhs]

@MachineLearningSebastian lol ik I'm just desperate lol I put raspberry pi os Debian 10 on my steam deck nothing ubuntu container was the closest I got but failed doing the test libedgetpu issue