cmake_opencv_fetchcontent

An example of using FetchContent to include OpenCV in your CMake C++ project

This VS Code project now uses a launch.json file, which enables us to define a sourceFileMap and step into OpenCV code while debugging. This means you should use the generic VS Code 'F5' Debug keyboard shortcut rather than the 'Ctrl+F5' provided by the CMake Extension.

Dependencies

• Install Visual Studio Build Tools 2022
  • or Visual Studio Community/Professional/Enterprise 2022
  • E.g. using one of the Evergreen Bootstrappers
• Install VS Code
• Install CMake
• Install the C/C++ Extension Pack in VS Code

Code/Build/Debug Cycle

• Open this folder in VS Code
• Press 'Ctrl+Shift+P'
• Execute the "CMake: Select a Kit" command
• Select the VS Build Tools 2022 amd64 kit
  • or Visual Studio Community/Professional/Enterprise 2022 amd64 kit
• Now press 'Ctrl+Shift+P' and do "CMake: Configure"
• Then 'Ctrl+Shift+P' and "CMake: Build"
• Press 'Ctrl+P' and open "main.cpp"
• Use F9 to place some breakpoint(s) in main()
• Press F5 to debug the example OpenCV program in main.cpp
• Try stepping into some OpenCV code using F11

OpenCV in the CERN ROOT C++ Interpreter

If you want to use OpenCV from within ROOT or cling, first make sure you uninstall (using the Visual Studio Installer) all Windows Kits (SDKs) except the one that ROOT was built with. At the time of writing, the latest ROOT release for windows was built with Windows SDK 10.0.19041.0.

Once you have built this CMake project you can start ROOT:

user@machine MSYS ~/Documents/GitHub/cmake_opencv_fetchcontent (main)
$ /c/root_v6.28.06/bin/root.exe
   ------------------------------------------------------------------
  | Welcome to ROOT 6.28/06                        https://root.cern |
  | (c) 1995-2023, The ROOT Team; conception: R. Brun, F. Rademakers |
  | Built for win64 on Aug 28 2023, 11:29:15                         |
  | From tags/v6-28-06@v6-28-06                                      |
  | With MSVC 19.32.31332.0                                          |
  | Try '.help'/'.?', '.demo', '.license', '.credits', '.quit'/'.q'  |
   ------------------------------------------------------------------

Now to get OpenCV into the interpreter:

root [] #pragma cling load("build/_deps/opencv-src/build/x64/vc16/bin/opencv_world480.dll")
root [] #pragma cling add_include_path("build/_deps/opencv-src/build/include")
root [] #include <opencv2/core.hpp>
root [] #include <opencv2/imgproc.hpp>
root [] #include <opencv2/highgui.hpp>
root [] using namespace cv;
root [] char wndname[] = "OpenCV Example";
root [] int width = 1024; int height = 768;
root [] Mat image = Mat::zeros(height, width, CV_8UC3);
root [] Size textsize = getTextSize("Hello OpenCV!", FONT_HERSHEY_COMPLEX, 3, 5, 0);
root [] Point org((width - textsize.width)/2, (height - textsize.height)/2);
root [] putText(image, "Hello OpenCV!", org, FONT_HERSHEY_COMPLEX, 3, Scalar(0, 0, 255), 5, LINE_AA);
root [] imshow(wndname, image);

At this point the "Hello, OpenCV!" window should be displayed, but unresponsive. Calling waitKey() enters the OpenCV highgui eventloop.

root [] waitKey();

Press any key to get out of the event loop and back to ROOT. To get rid of the lingering window:

root [] destroyAllWindows();

When you're done interpreting C++ code, type '.q'.

root [] .q
user@machine MSYS ~/Documents/GitHub/cmake_opencv_fetchcontent (main)