Main.py crashes with Arch-based distro, + workaround

[autodistries]

I have tried running FoxyFace on my laptop (which runs Garuda Linux, based on Arch) and it failed. It also failed on a brand-new Arch system, so this is either the hardware of my computer that's at fault, or any Arch-based distro will similarly fail. I have not tested other distros.

Laptop has an integrated Alder-Lake Intel GPU, and a discrete Nvidia gpu which is configured well enough. I use envycontrol to choose what GPU to use, and I only tested FoxyFace while both GPUs were available to use. I can run anything on my nvidia gpu (either with prime-run xxx, or with __GLX_VENDOR_PRELOADER=nvidia xxx. It shows up in nvidia-smi).

I have tried both with the python312 aur package and pyenv-managed python, with the same results.
The error, which showed in the latest.log and in terminal, was the following :

latest.log

I've made the file a little bit more readable. This is the gist of it:

Traceback (most recent call last):
File "/home/user/FoxyFace/FoxyFace/Main.py", line 30, in <module>
from src.UpdateChecker import UpdateChecker
File "/home/user/FoxyFace/FoxyFace/src/UpdateChecker.py", line 10, in <module>
from src.ui.windows.MainWindow import MainWindow
File "/home/user/FoxyFace/FoxyFace/src/ui/windows/MainWindow.py", line 12, in <module>
from src.pipline.BabblePipeline import BabblePipeline
File "/home/user/FoxyFace/FoxyFace/src/pipline/BabblePipeline.py", line 7, in <module>
from src.pipline.MediaPipePipeline import MediaPipePipeline
File "/home/user/FoxyFace/FoxyFace/src/pipline/MediaPipePipeline.py", line 17, in <module>
from src.stream.mediapipe.core.MediaPipeFrame import MediaPipeFrame
File "/home/user/FoxyFace/FoxyFace/src/stream/mediapipe/core/MediaPipeFrame.py", line 3, in <module>
from mediapipe.tasks.python.vision.face_landmarker import FaceLandmarkerResult
File "/home/user/.pyenv/versions/venv/lib/python3.12/site-packages/mediapipe/__init__.py", line 16, in <module>
import mediapipe.python.solutions as solutions 
File "/home/user/.pyenv/versions/venv/lib/python3.12/site-packages/mediapipe/python/solutions/__init__.py", line 17, in <module>
import mediapipe.python.solutions.drawing_styles
File "/home/user/.pyenv/versions/venv/lib/python3.12/site-packages/mediapipe/python/solutions/drawing_styles.py", line 21, in <module>
from mediapipe.python.solutions.hands import HandLandmark
File "/home/user/.pyenv/versions/venv/lib/python3.12/site-packages/mediapipe/python/solutions/hands.py", line 31, in <module>
from mediapipe.calculators.tflite import ssd_anchors_calculator_pb2
File "/home/user/.pyenv/versions/venv/lib/python3.12/site-packages/mediapipe/calculators/tflite/ssd_anchors_calculator_pb2.py", line 23, in <module>
DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n9mediapipe/calculators/tflite/ssd_anchors_calculator.proto\x12\tmediapipe\x1a$mediapipe/framework/calculator.proto\x1a\x39mediapipe/framework/formats/object_detection/anchor.proto\"\xd7\x05\n\x1bSsdAnchorsCalculatorOptions\x12\x18\n\x10input_size_width\x18\x01 \x01(\x05\x12\x19\n\x11input_size_height\x18\x02 \x01(\x05\x12\x11\n\tmin_scale\x18\x03 \x01(\x02\x12\x11\n\tmax_scale\x18\x04 \x01(\x02\x12\x1c\n\x0f\x61nchor_offset_x\x18\x05 \x01(\x02:\x03\x30.5\x12\x1c\n\x0f\x61nchor_offset_y\x18\x06 \x01(\x02:\x03\x30.5\x12\x12\n\nnum_layers\x18\x07 \x01(\x05\x12\x19\n\x11\x66\x65\x61ture_map_width\x18\x08 \x03(\x05\x12\x1a\n\x12\x66\x65\x61ture_map_height\x18\t \x03(\x05\x12\x0f\n\x07strides\x18\n \x03(\x05\x12\x15\n\raspect_ratios\x18\x0b \x03(\x02\x12+\n\x1creduce_boxes_in_lowest_layer\x18\x0c \x01(\x08:\x05\x66\x61lse\x12*\n\x1finterpolated_scale_aspect_ratio\x18\r \x01(\x02:\x01\x31\x12 \n\x11\x66ixed_anchor_size\x18\x0e \x01(\x08:\x05\x66\x61lse\x12+\n\x1cmultiscale_anchor_generation\x18\x0f \x01(\x08:\x05\x66\x61lse\x12\x14\n\tmin_level\x18\x10 \x01(\x05:\x01\x33\x12\x14\n\tmax_level\x18\x11 \x01(\x05:\x01\x37\x12\x17\n\x0c\x61nchor_scale\x18\x12 \x01(\x02:\x01\x34\x12\x1c\n\x11scales_per_octave\x18\x13 \x01(\x05:\x01\x32\x12#\n\x15normalize_coordinates\x18\x14 \x01(\x08:\x04true\x12(\n\rfixed_anchors\x18\x15 \x03(\x0b\x32\x11.mediapipe.Anchor2T\n\x03\x65xt\x12\x1c.mediapipe.CalculatorOptions\x18\xff\xb8\xf3u \x01(\x0b\x32&.mediapipe.SsdAnchorsCalculatorOptions')
TypeError : Couldn't build proto file into descriptor pool: Invalid default '0.5' for field mediapipe.SsdAnchorsCalculatorOptions.anchor_offset_x of type 2

I got around the problem by installing distrobox (in the extra repo). I created a debian-based box, entered it, installed pyenv, python3.12.9, libportaudio2, then created a virtualenv, activated it, installed the python FoxyFace requirements, then ran the app :

distrobox create --name debian-latest --image debian:latest
distrobox enter debian-latest # this drops you in the box's shell
bash --norc # The .bashrc/fishrc are shared. So if your host system has aliases like   apt=man pacman   this is necessary
sudo apt install pyenv
pyenv install 3.12.9 # pyenv uses your home dir to store stuff. So you may already have a correct python version installed. Check with : pyenv versions
eval "$(pyenv init - bash)" # This lets pyenv take control of what happens when you run python
cd FoxyFace/FoxyFace # your folders are shared between host and box
pyenv shell 3.12.9 # sets the default python version for this shell session to 3.12.9
python --version # make sure all is good
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python Main.py

I could not get it to run on my NVIDIA GPU. There are supposedly guides for this but I couldn't get it to work. I still get fairly low latency with my integrated cpu.
Anyway, I run VRCFT on my host machine and there are no issues for them to communicate.

I have created a .desktop that runs the following command for easier launching :

distrobox enter debian-latest -- /bin/bash --norc -c 'cd /path/to/FoxyFace/FoxyFace && eval "$(pyenv init - bash)" && source venv/bin/activate && python Main.py'

(I set it to open in a terminal, because closing the window seemingly doesn't fully exit the app)

Thanks for @Jeka8833 for their time and the help witg debugging c:

Edit : If I run the program with the discrete NVidia GPU disabled, I get *8 in image processing delay, so some obscure stuff is happening somewhere

[gitdunk]

Doesn't happen on my Endeavouros PC, b̶u̶t̶ ̶I̶ ̶a̶l̶s̶o̶ ̶d̶o̶n̶'̶t̶ ̶h̶a̶v̶e̶ ̶a̶n̶y̶ ̶o̶f̶ ̶t̶h̶e̶ ̶v̶i̶d̶e̶o̶ ̶c̶a̶r̶d̶ ̶c̶o̶m̶p̶o̶n̶e̶n̶t̶s̶ ̶i̶t̶ ̶s̶a̶y̶s̶ ̶y̶o̶u̶ ̶n̶e̶e̶d̶ ̶a̶n̶d̶ ̶i̶t̶ ̶s̶t̶i̶l̶l̶ ̶r̶u̶n̶s̶ ̶p̶e̶r̶f̶e̶c̶t̶l̶y̶ ̶s̶o̶ ̶s̶o̶m̶e̶t̶h̶i̶n̶g̶ ̶i̶s̶ ̶o̶b̶v̶i̶o̶u̶s̶l̶y̶ ̶w̶e̶i̶r̶d̶ ̶a̶b̶o̶u̶t̶ ̶m̶i̶n̶e̶ ̶(̶N̶V̶I̶D̶I̶A̶ ̶w̶i̶t̶h̶ ̶p̶r̶o̶p̶r̶i̶e̶t̶a̶r̶y̶ ̶d̶r̶i̶v̶e̶r̶s̶ ̶b̶u̶t̶ ̶n̶o̶t̶ ̶c̶u̶d̶a̶ ̶b̶t̶w̶ ̶i̶f̶ ̶t̶h̶a̶t̶ ̶i̶s̶ ̶i̶m̶p̶o̶r̶t̶a̶n̶t̶) Scratch that, the requirements.txt installed the onnx-runtime automatically.

[autodistries]

Doesn't happen on my Endeavouros PC
Ah, interesting. I'll install it and see if it gets solved

[BoheSama1999]

It works on my ArchPC on kdewayland (kernel 6.17.9-zen1-1-zen, nvidia-open-dkms driver, cuda 12.8.1-3, cudnn 9.11.0.98-3)

(FoxyFace) [NekoType@NekoArch FoxyFace]$ uv pip list
Using Python 3.12.12 environment at: FoxyFace
Package                  Version
------------------------ ------------
absl-py                  2.3.1
attrs                    25.4.0
certifi                  2025.11.12
cffi                     2.0.0
charset-normalizer       3.4.4
coloredlogs              15.0.1
contourpy                1.3.3
cycler                   0.12.1
dataclass-wizard         0.35.1
filelock                 3.19.1
flatbuffers              25.9.23
fonttools                4.61.0
fsspec                   2025.9.0
humanfriendly            10.0
idna                     3.11
jax                      0.7.1
jaxlib                   0.7.1
jinja2                   3.1.6
kiwisolver               1.4.9
markupsafe               2.1.5
matplotlib               3.10.7
mediapipe                0.10.21
ml-dtypes                0.5.4
mpmath                   1.3.0
networkx                 3.5
numpy                    1.26.4
nvidia-cublas            13.0.0.19
nvidia-cublas-cu12       12.8.4.1
nvidia-cuda-cupti        13.0.48
nvidia-cuda-cupti-cu12   12.8.90
nvidia-cuda-nvrtc        13.0.48
nvidia-cuda-nvrtc-cu12   12.8.93
nvidia-cuda-runtime      13.0.48
nvidia-cuda-runtime-cu12 12.8.90
nvidia-cudnn-cu12        9.10.2.21
nvidia-cudnn-cu13        9.13.0.50
nvidia-cufft             12.0.0.15
nvidia-cufft-cu12        11.3.3.83
nvidia-cufile            1.15.0.42
nvidia-cufile-cu12       1.13.1.3
nvidia-curand            10.4.0.35
nvidia-curand-cu12       10.3.9.90
nvidia-cusolver          12.0.3.29
nvidia-cusolver-cu12     11.7.3.90
nvidia-cusparse          12.6.2.49
nvidia-cusparse-cu12     12.5.8.93
nvidia-cusparselt-cu12   0.7.1
nvidia-cusparselt-cu13   0.8.0
nvidia-nccl-cu12         2.27.5
nvidia-nccl-cu13         2.27.7
nvidia-nvjitlink         13.0.39
nvidia-nvjitlink-cu12    12.8.93
nvidia-nvshmem-cu12      3.3.20
nvidia-nvshmem-cu13      3.3.24
nvidia-nvtx              13.0.39
nvidia-nvtx-cu12         12.8.90
oneeurofilter            0.2.1
onnxruntime-gpu          1.23.2
opencv-contrib-python    4.11.0.86
opt-einsum               3.4.0
packaging                25.0
pillow                   12.0.0
protobuf                 4.25.8
pycparser                2.23
pyparsing                3.2.5
pyside6                  6.10.1
pyside6-addons           6.10.1
pyside6-essentials       6.10.1
python-dateutil          2.9.0.post0
requests                 2.32.5
scipy                    1.16.3
sentencepiece            0.2.1
setuptools               70.2.0
shiboken6                6.10.1
six                      1.17.0
sounddevice              0.5.3
soundfile                0.13.1
sympy                    1.14.0
torch                    2.9.1+cu128
torchaudio               2.9.1+cu128
torchvision              0.24.1+cu128
triton                   3.5.1
typing-extensions        4.15.0
tzdata                   2025.2
urllib3                  2.6.0

[BoheSama1999]

My steps:
git clone --recurse-submodules https://github.com/Jeka8833/FoxyFace.git

then goto FoxyFace/FoxyFace create venv
uv venv FoxyFace --python 3.12

activate the venv
source FoxyFace/bin/activate

uv pip install -r requirements.txt

install cuda 12.8 from archive.archlinux.org
sudo pacman -U https://archive.archlinux.org/packages/c/cuda/cuda-12.8.1-3-x86_64.pkg.tar.zst

install cudnn 9.11.0 from archive.archlinux.org, after 9.11.0 required cuda13 but onnxruntime not supported cuda13 now
sudo pacman -U https://archive.archlinux.org/packages/c/cudnn/cudnn-9.11.0.98-3-x86_64.pkg.tar.zst

install pytorch
uv pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu128

install onnxruntime-gpu
uv pip install onnxruntime-gpu

Last run Main.py
python Main.py

[Jeka8833]

I will most likely try to make corrections to my documentation, as you have provided useful information. Although I have no idea why we would need to use torchaudio.

install cuda 12.8 from archive.archlinux.org
sudo pacman -U https://archive.archlinux.org/packages/c/cuda/cuda-12.8.1-3-x86_64.pkg.tar.zst

install cudnn 9.11.0 from archive.archlinux.org, after 9.11.0 required cuda13 but onnxruntime not supported cuda13 now
sudo pacman -U https://archive.archlinux.org/packages/c/cudnn/cudnn-9.11.0.98-3-x86_64.pkg.tar.zst

install pytorch
uv pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu128

[BoheSama1999]

I will most likely try to make corrections to my documentation, as you have provided useful information. Although I have no idea why we would need to use torchaudio.

install cuda 12.8 from archive.archlinux.org
sudo pacman -U https://archive.archlinux.org/packages/c/cuda/cuda-12.8.1-3-x86_64.pkg.tar.zst
install cudnn 9.11.0 from archive.archlinux.org, after 9.11.0 required cuda13 but onnxruntime not supported cuda13 now
sudo pacman -U https://archive.archlinux.org/packages/c/cudnn/cudnn-9.11.0.98-3-x86_64.pkg.tar.zst
install pytorch
uv pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu128

oh sure torchaudio is redundant, I'm just in the habit of

[autodistries]

funnily enough, I did not manage to replicate the fix. I'll try on a clean arch install and update this comment

[StyledStrike]

funnily enough, I did not manage to replicate the fix. I'll try on a clean arch install and update this comment

Same here, the same error shows up even after using BoheSama's steps

[StyledStrike]

Update: I forgot to mention I was using CachyOS. After switching to regular Arch Linux, the regular installation guide worked just fine