Can I run my own custom code on the Hailo AI HAT+ instead of using pre-built models?

General
1
  1. Is it possible to use two cameras simultaneously for object detection with the Hailo AI HAT+?
    2/ Can I run my own custom code on the Hailo AI HAT+ instead of using pre-built models? If so, what are the recommended steps for deploying custom AI models or scripts?
2

Welcome to the Hailo Community!

That is up to your application. The Hailo-8/8L accelerator can run multiple models and can run inference for multiple cameras.

Yes, you can run your own models. Start by downloading the Hailo AI Software Suite Docker and run the tutorials to find out how to convert models. Inside the Docker type:

hailo tutorials

This will start a Jupyter notebook server with notebooks for each step of the model conversion process.

3

I’m trying to run object detection using two cameras simultaneously with the Hailo AI HAT+. However, I’m currently stuck using basic pipelines and can only set --input rpi, which limits me to just one camera.

I have already successfully converted my model, but I can’t get the code to run with multiple cameras due to this restriction in the basic pipeline. Could you provide guidance on how to modify the pipeline to support multiple camera inputs?

Thanks in advance for your help!

4

Hi @Anuj_Ch
You can see this topic for some example: Running multiple models independently

5

Here is another thread that solved display issue on raspberry pis: Using gstreamer with 2 different USB camera inputs and one .hef model - General - Hailo Community