Run PaddleOcr recognition model in Hailo AI accelerator

Guides
1

Hi,
I’m trying to run the PaddleOCR recognition model on a Raspberry Pi using the Hailo AI Accelerator. I would like to know the steps or procedures needed to:

Convert the PaddleOCR recognition model to be compatible with Hailo,
And perform inference on the Raspberry Pi using the Hailo device.

Could you guide me on the required tools, supported model formats, conversion pipeline, and inference setup? Any help or reference documentation would be appreciated.

2

Hi @Rithiga_S!

Welcome to the Hailo Community! Great to have you here.

So for what you’re trying to do, here’s the general approach:

First, you’ll need to get your model in ONNX format and then compile it for Hailo using either the Hailo Model Zoo or the Hailo DFC (Dataflow Compiler). Once that’s done, you can deploy and use it with the Raspberry Pi examples we provide.

We actually have a license plate recognition example that might be helpful: tappas/apps/h8/gstreamer/general/license_plate_recognition at master · hailo-ai/tappas · GitHub

Just a heads up though - it’s not updated to our latest version and it’s specifically designed for Israeli license plates, so you might need to adapt it for your use case.

The good news is we’re working on a new, updated example that will be added to our Raspberry Pi 5 examples repository: GitHub - hailo-ai/hailo-rpi5-examples

Hope this helps get you started! Let me know if you have any other questions.

3

Hi @omria,

Thanks or your response.