@JonB Thanks, I normally use CQTDeployer. That would mean for other projects the DLLs also need to be added under LIBS in the .pro file ?

@BigBen
This sounds like you would be best looking for openCV's VideoCapture.open() questions? Maybe whatever it does requires certain files in certain places which are not present/found on the other system, either openCV runtime files or the file you are trying to open?

Hi and welcome to devnet,

Are you sure you are linking all the required libraries ?
Are they all from compatible compilers ?

@BigBen
Yes, you always need to write a new class which derives from the base class if you need to override a virtual method, as QWidget::paintEvent().

The Analog Clock Example illustrates overriding paintEvent(). There will be hundreds of other examples online.

I am wondering whether there's some bad interaction between the 2017 build of Qt and your 2019 build of OpenCV. There should not as VS 2019 is backward compatible with 2017.

Hi and welcome to devnet,

You can't jump from a 32bit float image to 16bit integer one like that.

Convert your OpenCV image to 16 bit before making a QImage out of it.

Hi,

You should check your video file, but as I wild guess, it's likely a FullHD video which is 1920 * 1024. Since you are creating a RGB QImage out of it, it's already weighing about 6MB per image. QPixmap is meant to be optimised for rendering and thus the backing memory used might even be larger for, for example, alpha channel handling. Depending on what else you do with your application and these images you can pretty quickly hit memory limits if your application is 32bit.

Lack INCLUDEPATH such as:
INCLUDEPATH += /usr/local/include

Thank you all guys. Your suggestions and expert opinion helped a lot with solving the problem. QT forum is always so helpful.

After trying many things here is the conclusion and code is now working perfectly fine

I removed the tic toc part

time(&start);
timer = double(getTickCount());
tic();
It was working but crashing then just to make sure that QImage is not NULL I removed the img = QImage(); // releasing memory

with

if(!img.isNull())
img = QImage();
Its working perfectly fine now.

@Imran-Hassan said in QT image display is not working without declaring openCV namedWindow in windows 10:

The problem is solved

great, so please don't forget to mark your post as such!

I still don't know what exactly went wrong, but after I removed all of the OpenCV stuff and re-installed a fresh v4.1, it worked. So even though I checked all lib-files, there must have been something wrong with the linked libs (Pointing to different versions?! Links not working?!, etc.)

Thanks @SGaist, your answer made me remove everthing and try the OCV version that is provided by Ubuntu 18.04. I got 3.X.X from there... After I removed this version again (because I want a newer one) and installed 4.1 from OCV page, it worked :-)

@jsulm Hello! Sorry, but that doesn't work. I tried to insert a loop in different places, but the result remains the same. I also tried to display the main widget.
Снимок экрана от 2019-11-27 08-27-43.png

@SGaist Hello. So the VideoWindow.h file (VideoWindow class) is my custom widget. This is the first piece of code that I presented.

@jsulm said in QDataStream an openCV::mat serialisation...:

@Dariusz What is stream here? Is it QDataStream?
"Any idea what did I mess up with this ?" - well, what is not working?

Yep :- )

I think I might "got it". Naturally 5 min after asking question... elemSize1() return incorrect number of channels. Looks like I need elemSize() instead. still testing.

Yep it was wrong element number.

Hope it helps other "serializers" out here : -)

@r-p-h said in OpenCV 4.1.1 Imread() & Imwrite() Crashes Program ?:

now everything is working fine.

so please don't forget to mark your post as solved! Thanks.

An update on this. It works perfectly and its a very low-cost solution. My processing time remains almost the same (~10mS) using this method since the painter does the scaling for you when it draws.

Thanks for your help guys.

@chris1092387456 could it be possible you have mixed Qt versions from the one you cross-compiled and the one it's used by the OpenCV install script you used.

Assuming you run Ubuntu/opencv_latest.sh which in turn calls opencv_install.sh and then dependencies.sh, there are some apt-get install calls, so you''ll end up having pre-built Qt libraries along with your cross-compiled ones...

function install_dependency { echo "--- Installing dependency: $1" sudo apt-get -y install $1 } ... install_dependency libqt5x11extras5 install_dependency libqt5opengl5 install_dependency libqt5opengl5-dev

So may I suggest to start from scratch?

First I'd follow the mechatronics guide to cross-compile and deploy Qt in the RPi device.

Then I'd follow this guide to cross compile the latest version of OpenCV for Raspberry Pi. As you'll use WITH_QT=ON flag I guess, please adjust the paths in the CMake script(s) to use the cross-compiled Qt version in your host PC.

And also use the correct path for cross-compiled OpenCV in your .pro file. As example, the following doesn't look good to me:

LIBS += -L/usr/local/lib/libopencv_highgui.so

@Gojir4 thanks, I am trying that right now. I’ve added a 1s delay in my image grabbing thread. Let’s see. Fingers crossed!