Hailo-10H Runtime Setup and CV Model Deployment

Use this guide to install and verify the Hailo-10H runtime and deploy an AI Lab CV model on a Raspberry Pi CM5 with a PCIe Hailo-10H accelerator. This page covers the common device, runtime, and container workflow; model-specific arguments remain on each model card. Hailo-10H uses its own HEF, image, Python binding, and HailoRT library; none may be mixed with Hailo-8 artifacts.

1. Install and Verify the Runtime

On a supported Raspberry Pi OS image, install the Hailo-10H host package and restart before first use:

bash

sudo apt update
sudo apt install hailo-h10-all -y
sudo reboot

After reconnecting, verify the hardware:

bash

lspci -nn | grep -i hailo
lsmod | grep hailo
hailortcli --version
ls -l /dev/hailo0
sudo hailortcli fw-control identify

Expected result: the PCIe device is identified as Hailo-10H and /dev/hailo0 exists. If it does not, resolve driver, PCIe, or power issues before running a model container.

2. Pull the Model Image

Use the exact Hailo-10H image shown on the AI Lab model card:

bash

sudo docker pull ghcr.io/seeed-projects/recomputer-hailo10h-cv/<model>:latest

3. Run the Model

Replace placeholders with the model card command. The versioned HailoRT shared library and the unversioned symlink must both be mounted.

bash

sudo docker run --rm \
  --name cm5-hailo10h-<model> \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  -v /usr/lib/libhailort.so.5.1.1:/usr/lib/libhailort.so.5.1.1:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo10h-cv/<model>:latest \
  python web_detection.py --model_path model/<model>.hef --video_path video/test.mp4

Open http://<BOARD_IP>:8000 when the service reports that the web preview is ready. Keep any model-specific arguments from the card, including secondary HEFs used by two-stage pipelines.

4. Diagnose Common Failures

Symptom	Check
`/dev/hailo0` does not exist	Confirm `lspci` sees Hailo-10H, the driver is loaded, and restart after installation.
`Failed to open device file`	Check device permissions, the container `--device` argument, and HailoRT host setup.
HailoRT Python or shared-library error	Use the Hailo-10H image and mount both `libhailort.so.5.1.1` and `libhailort.so`.
Inference API error	Confirm the image tag includes the fixed application code and the HEF is Hailo-10H-specific.

5. Before Publishing a New Model

Validate the image and command on physical Hailo-10H hardware. Then add the image package as a downloadSources entry in AI Lab and create a dedicated Hailo-10H Wiki so users receive the correct command after choosing the platform.