fresh rpi5 with ai hat+ (26) and first try to run a cv test app fails.

General Discussion
,
1

Hi

Just bought a rpi5 with ai hat+(26) for some personal trials on cv. I did some cv development on earlier rpi models and it was ok then. Now with this new toy I’m expecting a great leap for my development tests.

Unfortunately t took me two days to properly configure the hardware, stock images didn’t help and finding the right info and package is a little but confusing. Anyway I visited all possible methods (deb packages, apt repositories, compile from source, gits, forums, ai agents etc) and now I have a working hardware with 6.12 kernel, 4.23 hailort and 3.12 python. I’m using a headless setup with remote vscode dev environment. I started with a basic cv code I’ve been working on on another platform. I modified the code to include hailort support for processing. And before my first run when I install the requirements on venv via pip, I encountered a strange message where hailort requires numpy < 2 where opencv install 2.2. My guess I’m on the wrong side of the dependency wars. I just need somebody to point me the right direction where I can focus on my code instead of this holy wars. Any help is appreciated. Thanks…

Here is the description of my current situation:

rpi5dev@rpi5-dev:~/silX $ hailortcli fw-control identify
Executing on device: 0001:01:00.0
Identifying board
Control Protocol Version: 2
Firmware Version: 4.23.0 (release,app,extended context switch buffer)
Logger Version: 0
Board Name: Hailo-8
Device Architecture: HAILO8

rpi5dev@rpi5-dev:~/silX $ source venv/bin/activate
((venv) ) rpi5dev@rpi5-dev:~/silX $ cd age-genderX/
((venv) ) rpi5dev@rpi5-dev:~/silX/age-genderX $ pip install -r requirements.txt
Collecting opencv-python-headless (from -r requirements.txt (line 1))
Using cached opencv_python_headless-4.12.0.88-cp37-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (19 kB)
Requirement already satisfied: numpy in /home/rpi5dev/silX/venv/lib/python3.12/site-packages (from -r requirements.txt (line 2)) (1.26.4)
Collecting numpy (from -r requirements.txt (line 2))
Using cached numpy-2.2.6-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl.metadata (63 kB)
Using cached opencv_python_headless-4.12.0.88-cp37-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (33.1 MB)
Using cached numpy-2.2.6-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (14.0 MB)
Installing collected packages: numpy, opencv-python-headless
Attempting uninstall: numpy
Found existing installation: numpy 1.26.4
Uninstalling numpy-1.26.4:
Successfully uninstalled numpy-1.26.4
ERROR: pip’s dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
hailort 4.23.0 requires numpy<2, but you have numpy 2.2.6 which is incompatible.
Successfully installed numpy-2.2.6 opencv-python-headless-4.12.0.88

[notice] A new release of pip is available: 25.0.1 → 25.3
[notice] To update, run: pip install --upgrade pip
((venv) ) rpi5dev@rpi5-dev:~/silX/age-genderX $ python3.12 console_app.py
:hammer_and_wrench: Model yükleniyor: models/face_attr.hef …
:information_source: Model Shape Raw: (218, 178, 3)
:white_check_mark: Model Hazır! Giriş Boyutu: 178x218
:cross_mark: KRİTİK HATA: Model yüklenemedi.
Detay: ‘ConfiguredNetwork’ object has no attribute ‘create_params_dict_by_name_with_group_params’
((venv) ) rpi5dev@rpi5-dev:~/silX/age-genderX $

2

Hey @AyhanGENC,

Welcome to the Hailo Community!

We actually got this sorted out with the 4.23 deb release for RPi—specifically the Trixie version with Python 3.13 support. That should be your go-to. I’d recommend running through our automatic installer to get everything set up properly:

sudo apt update
sudo apt full-upgrade
sudo apt install dkms
sudo apt install hailo-all

That should clear up the dependency issues you’re running into, especially if you’re on RPi5 with Trixie and Python 3.13. RPI apt server has the updated HailoRT package for RPi with NumPy 2+ support, so this will get you on the right version.

Hope this helps!

3

Hi @omria

Thanks for the tips. I’ll try that path as well. I managed to run the whole system with python 3.10 and hailort 4.18 though. With some trial I’ll decide if newer is better of mature is the choice for my specific needs. Great to be part of this community :blush:

1 Like