This repository provides a method of using the Raspberry Pi AI Kit on Ubuntu 24.04. It makes use of a Docker container to bundle the correct versions of the Hailo SDK libraries.
For the SDK to work with the AI accelerator hardware, a matching version of the driver needs to be used. Even a minor version difference will prevent the SDK from detecting the hardware.
Install requirements on host:
sudo apt-get install linux-headers-$(uname -r)
We get the exact version of the driver's source code from Github:
git clone https://github.com/hailo-ai/hailort-drivers.git
cd hailort-drivers
git checkout f840b6219230ec9a350444dbb903adbf0f63a373
Then build it and install it on the host system:
cd linux/pcie
make all
sudo make install
sudo modprobe hailo_pci
cd ../..
./download_firmware.sh
sudo mkdir -p /lib/firmware/hailo
sudo mv hailo8_fw.4.17.0.bin /lib/firmware/hailo/hailo8_fw.bin
sudo cp ./linux/pcie/51-hailo-udev.rules /etc/udev/rules.d/
sudo udevadm control --reload-rules && sudo udevadm trigger
It's better to restart after installing this driver. After a reboot you can look at the kernel buffer to see if the device is detected and the driver loaded.
$ sudo dmesg | grep hailo
...
[ 4.379687] hailo: Init module. driver version 4.17.0
...
[ 4.545602] hailo 0000:01:00.0: Firmware was loaded successfully
[ 4.572371] hailo 0000:01:00.0: Probing: Added board 1e60-2864, /dev/hailo0
Configure host X server to accept connections from a Docker container
xhost +local:docker
Build and start the container
docker compose build
docker compose up -d hailo-ubuntu-pi
Open a shell inside the container
docker compose exec hailo-ubuntu-pi /bin/bash
Test that the camera is working. On Ubuntu you will have to add the --qt-preview argument to all rpicam- commands.
This is because the default windowing toolkit used by these examples is not supported on Ubuntu.
rpicam-hello --timeout=0 --qt-preview
Install Hailo libraries. This can't be done as part of building the container, as configuring the packages depend on a running systemd.
apt install hailo-all
Check that the hardware is working from inside the container
$ hailortcli fw-control identify
Executing on device: 0000:01:00.0
Identifying board
Control Protocol Version: 2
Firmware Version: 4.17.0 (release,app,extended context switch buffer)
Logger Version: 0
Board Name: Hailo-8
Device Architecture: HAILO8L
Serial Number: <redacted>
Part Number: <redacted>
Product Name: HAILO-8L AI ACC M.2 B+M KEY MODULE EXT TMP
Test TAPPAS Core installation:
$ gst-inspect-1.0 hailotools
Plugin Details:
Name hailotools
Description hailo tools plugin
Filename /lib/aarch64-linux-gnu/gstreamer-1.0/libgsthailotools.so
Version 3.28.2
License unknown
Source module gst-hailo-tools
Binary package gst-hailo-tools
Origin URL https://hailo.ai/
hailoaggregator: hailoaggregator - Cascading
hailocounter: hailocounter - postprocessing element
hailocropper: hailocropper
hailoexportfile: hailoexportfile - export element
hailoexportzmq: hailoexportzmq - export element
hailofilter: hailofilter - postprocessing element
hailogallery: Hailo gallery element
hailograytonv12: hailograytonv12 - postprocessing element
hailoimportzmq: hailoimportzmq - import element
hailomuxer: Muxer pipeline merging
hailonv12togray: hailonv12togray - postprocessing element
hailonvalve: HailoNValve element
hailooverlay: hailooverlay - overlay element
hailoroundrobin: Input Round Robin element
hailostreamrouter: Hailo Stream Router
hailotileaggregator: hailotileaggregator
hailotilecropper: hailotilecropper - Tiling
hailotracker: Hailo object tracking element
18 features:
+-- 18 elements
$ gst-inspect-1.0 hailo
Plugin Details:
Name hailo
Description hailo gstreamer plugin
Filename /lib/aarch64-linux-gnu/gstreamer-1.0/libgsthailo.so
Version 1.0
License unknown
Source module hailo
Binary package GStreamer
Origin URL http://gstreamer.net/
hailodevicestats: hailodevicestats element
hailonet: hailonet element
synchailonet: sync hailonet element
3 features:
+-- 3 elements
For the Global Shutter camera one needs to specify the image size, using the --width and --height arguments.
rpicam-vid -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov8_inference.json --lores-width 640 --lores-height 640 --width 1456 --height 1088 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov6_inference.json --lores-width 640 --lores-height 640 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov8_inference.json --lores-width 640 --lores-height 640 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolox_inference.json --lores-width 640 --lores-height 640 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov5_personface.json --lores-width 640 --lores-height 640 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov5_segmentation.json --lores-width 640 --lores-height 640 --framerate 20 --qt-preview
rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov8_pose.json --lores-width 640 --lores-height 640 --qt-preview
Hailo publishes more examples for the Raspberry Pi 5. We can run these too.
cd hailo-rpi5-examples
source setup_env.sh
pip install -r requirements.txt
./compile_postprocess.sh
python basic_pipelines/detection.py --input resources/detection0.mp4
python basic_pipelines/detection.py -i rpi
python basic_pipelines/detection.py --labels-json resources/barcode-labels.json --hef resources/yolov8s-hailo8l-barcode.hef -i rpi
python basic_pipelines/pose_estimation.py -i rpi
python basic_pipelines/instance_segmentation.py -i rpi
python basic_pipelines/detection.py -i /dev/video8
If the Advanced Examples fail with the following error:
AfMode not supported by this camera, please retry with 'auto-focus-mode=AfModeManual'
Run these commands to change auto-focus-mode to manual:
sed -i -e 's/auto-focus-mode=2/auto-focus-mode=0/g' basic_pipelines/detection.py
sed -i -e 's/auto-focus-mode=2/auto-focus-mode=0/g' basic_pipelines/pose_estimation.py
sed -i -e 's/auto-focus-mode=2/auto-focus-mode=0/g' basic_pipelines/instance_segmentation.py