How to connect Hailo-8l Rpi-5 AI kit to Hailo Software Suite running in Windows 11 Docker

General
1

I have the Hailo Software Suite 2024-07 running in Docker with WSL Ubuntu 22.04 backend. And the Hailo software running on the Raspberry Pi AI Kit.
I’m not able to figure out how to connect these two systems using Ethernet cable as its mentioned either PCIe or Ethernet should work.
Kindly explain the steps as I want to develop/explore models in Docker Studio and then port it to RPi5. Thanks.

1 Like
2

Hey @Sarvesh,

Welcome to the Hailo Community!

If you’re aiming to run models on an RPI, there’s no need to connect them directly. Simply train your model on your PC, then transfer the app and HEF file to the RPI.

On your PC, compile the models using the DFC, and then run them on the RPI using HailoRT or the CLI.

Best Regards

2 Likes
3

Thanks @omria

I’m still learning the ropes. If I wish to see the Hailo device when running hailortcli scan from Docker, is there any way I can use ethernet to connect to RPi5. At present it says “Hailo devices not found”.
Just want to make sure of my understanding, if I connect Docker running laptop and RPi5 with ethernet cable can show the Hailo device in scan.
I also tried installing HailoRT on Windows 11 host and installed HailoRT deb file in Ubuntu before I ran the script: ./hailo_ai_sw_suite_docker_run.sh --hailort-enable-service
I may be mixing up things here so please bear with me. Thanks.

4

Litte more context to add. I intend to tune the LPR model in zoo and see the inference in real time on Hailo-8L. Or explore the AI studio process in general.
But removing it from RPI AI kit and then inserting it in my PCIe slot of laptop is a long process (if I remove WLAN card and replace it with Hailo on the Pcie slot).

5

Hey @sarvesh,

Thanks for the context.

You don’t need to remove the device to check it or anything like that. Here’s how we recommend working with the RPI AI kit:

  1. Install Hailo Packages: Follow the guide to install all the Hailo packages: Getting Started with RPI5 Hailo8L
  2. Model Changes: If you want to make changes to the model, you can install Docker or the DFC only on your PC. This doesn’t require a Hailo device to run; it’s solely for creating the model.
  3. App Creation: Develop the app that uses the Hailo chip, model, and software (this is typically done on the PC). For app examples, check out:
  1. Monitor RPI: If your RPI is running and not connected to any monitor, you can simply SSH into it to see it running.

Please feel free to ask any questions.

Best regards.

1 Like
6

I have a question regarding app creation. Where do you create apps? In VSCode or docker container? On the pc? And importing it to the rpi when finished? I‘m fairly new to this whole AI thing and been playing a bit but i‘m confused :sweat_smile:

And, what is possible with the RPI AI Kit?

I see many examples for video stuff, but can you run SMLs in hef with it and create a chatbot?

I would love to know some other use cases where the AI kit can be used and what the limits are. I saw you are working on running LLMs on the Hailo8L, is there an estimated time for this feature?

Thanks a lot in advance! :slight_smile: