Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found"

[michwil99]

@Hikari-Tretaki thank you very much!!

[yashwant]

Hi, The error still persists even when I use "pip install opencv-python-headless"

Error message:
qt.qpa.xcb: could not connect to display
qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in "" even though it was found.
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: eglfs, minimal, minimalegl, offscreen, vnc, webgl, xcb.

/opt/conda/envs/yashenv2/.tmp8gb_phfb: line 3: 17 Aborted python TAD_GUI_Contoller.py
ERROR conda.cli.main_run:execute(33): Subprocess for 'conda run ['python', 'TAD_GUI_Contoller.py']' command failed. (See above for error)

Other info: When I used opencv-python (Error was similar like above except it had - Could not load the Qt platform plugin "xcb" in ".../python3.7/site-packages/cv2/qt/plugins" even though it was found.). I am using docker (with Ubuntu 18 + WSL2). The problem persist only in docker container. (The python codes work fine in windows)

Am I missing some Ubuntu libraries? I tried installing libxcb's and many other libraries. It just does not work. Or is it related to PyQt5?

[jsulm]

@yashwant said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

Am I missing some Ubuntu libraries?

Yes.
Set QT_DEBUG_PLUGINS before starting the app and check output.

[yashwant]

@jsulm said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

Set QT_DEBUG_PLUGINS

Can you tell me how can I set "export QT_DEBUG_PLUGINS =1" in docker container? Should it be written inside Dockerfile? or should I run the container something like docker run -it Image -v export QT_DEBUG_PLUGINS =1 ? or should I be using docker exec ?

[jsulm]

@yashwant That I don't know, sorry.

[yashwant]

ok :( any thanks!

[Pablo J. Rogina]

@yashwant said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

qt.qpa.xcb: could not connect to display

Are you sure you have an X server running and available?

I am using docker (with Ubuntu 18 + WSL2). The problem persist only in docker container.

Please be aware that you have a (very) unusual environment.

Given that Docker is involved you may need to check how to make the container output work properly under X...

[yashwant]

@Pablo-J-Rogina said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

rly under

1. It is just standard docker windows setup (It just uses the WSL2). It is the standard installation process.
2. How can I check if the Xserver is running?

Thanks !

[yashwant]

I am just trying to open a python GUI(which opens fine in windows 10) using docker setup. I did install docker for windows using the standard procedure. And I can load the python codes (without .ui - python files without GUI) into the container without any problem. The problem arises only when I try to load python files with GUI.

[Pablo J. Rogina]

@yashwant said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

The problem arises only when I try to load python files with GUI.

That's the point. You may need to check how to make a Docker container use the GUI of the host...

[alphaomega]

Replacing opencv-python with opencv-python-headless is one option. However, for some reasons if you want to keep the library, then you could add following lines after import cv2:

import os, sys
ci_build_and_not_headless = False
try:
    from cv2.version import ci_build, headless
    ci_and_not_headless = ci_build and not headless
except:
    pass
if sys.platform.startswith("linux") and ci_and_not_headless:
    os.environ.pop("QT_QPA_PLATFORM_PLUGIN_PATH")
if sys.platform.startswith("linux") and ci_and_not_headless:
    os.environ.pop("QT_QPA_FONTDIR")

This is based on the cv2's init file . What it does is essentially removing environment variables that are added by the init file (This are only done at runtime and does not have permanent effect upon your system.)

[yashwant]

@Pablo-J-Rogina said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

UI of the host..

Yes that was my goal (docker with gui)! and above mentioned was my error when I try GUI on docker.! Anyway thanks!

[Pablo J. Rogina]

@yashwant said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

that was my goal (docker with gui)!

Yes, I understand that. Please take into account that this issue is not related (or produced by) Qt itself so the solution might come from some other source(s), not only this forum

[redeyed]

@jeremy_k said in Using PyQt5 with opencv-python (cv2) causes error "could not load Qt platform plugin xcb even though it was found":

worka

Hello guys, so the problem is that import cv2 sets QT_* env vars.
I came to this workaround, maybe it will help someone:

    import cv2
    for k, v in os.environ.items():
        if k.startswith("QT_") and "cv2" in v:
            del os.environ[k]

[OBaladi]

I solved the problem by downgrading the cv2 version, simply run
python3 -m pip install opencv-contrib-python==4.1.0.25
i used to have the 4.5.3.56

[eyllanesc]

@lvlanson Also see https://stackoverflow.com/a/68417901/6622587

[Thiago Oliveira]

@Hikari-Tretaki, muito obrigado por sua ajuda. Também consegui resolver o meu problema com sua dica.

[aria212019005]

@redeyed Thankyou, I was confused around installing

opencv-python

and

opencv-python-headless

before this. But it ends with "we need both libraries, and this"

[Vijay Kumar]

@Hikari-Tretaki Thank you so much, installing opencv-python-headless fixed the issue

[Vitor Dalla Vecchia]

@Hikari-Tretaki MAN, I LOVE YOU SO MUCH. THANKS!