Custom yolov8 model on Raspberry pi and AI kit dont work

General
1

Hi all,

I’m trying to deploy my retrained YOLOv8n model on a Raspberry Pi 5 with the Hailo AI HAT kit.

I’m using the basic detect.py pipeline to load the labels and the compiled HEF file. The script runs without any errors in the terminal, but the video window doesn’t appear, and I can’t exit the process with Ctrl+C — it just hangs.

Has anyone experienced this before or knows what might be causing it?

Thanks in advance!

2

Hey @Henry_Huang,

Can you share the output of the terminal or the log so we can help you better ?

3

Thank you for your reply omria, please see below:

image
image886×722 126 KB

it just stuck there, no error and no respond.

4

oh also, below is my json file:

image
image327×222 9.86 KB

5

furthermore I took a screenshot when finish the compilng shiw below not sure if that error will be reason:

compile result
compile result1850×1055 221 KB

6

Hey @Henry_Huang,

Thanks for the info!

A few things to check:

  1. Can you see video/camera output when running the default detection? Just want to make sure the basic pipeline is working.

  2. If you only provide the HEF file (without the JSON), does it still get stuck? This would help isolate whether it’s a model issue or config issue.

  3. Can you run these commands to help debug:

    First, let’s make sure everything’s working at the hardware level:

    hailortcli fw-control identify

    Then test the HEF file directly:

    hailortcli run {your_hef_file}

    If those two work correctly, try running your detection from video or USB camera with the same HEF and JSON files.

This should help us figure out where exactly things are getting stuck. Let me know what you find!