KhuCv

• KhuCv: Open development source for computer vision (Kyung Hee University, Data Analysis & Vision Intelligence)
• C++ (wxWidgets + OpenCV), C++17

Introduction

• Open program sources
• wxWidgets 3.x + OpenCV 4.x
• Files open by OpenCV
• Files save by OpenCV (individual image)
• Images can be displayed in any position and can be moved by mouse
• Messages using format specifiers can be written to the output window (list box)
• Sequence processing (webcam, image sequence, video file)

Environment setting

• wxWidgets 3.x

  • Windows (+Visual Studio)
    • Download wxWidgets source code from https://www.wxwidgets.org/downloads/
    • Open wxWidgets-x.x.x/build/msw/wx_vcxx.sin with Visual Studio and build (release and debug)
  • Linux (Ubuntu)
    • Install g++-13(Ubuntu 20.04 or below versions, see https://lindevs.com/install-g-on-ubuntu/)
    • Download wxWidgets source code from https://www.wxwidgets.org/downloads/
    • $ sudo apt install libgtk-3-dev
    • $ mkdir build-gtk
    • $ cd build-gtk
    • $ ../configure --with-gtk
    • $ make
    • $ sudo make install
    • $ sudo ldconfig
    • $ cd samples
    • $ make
    • $cd ../demos
    • $make
  • Mac
    • Download wxWidgets source code from https://www.wxwidgets.org/downloads/
    • $ mkdir build-cocoa
    • [$ mkdir build-cocoa-debug]
    • $ cd build-cocoa
    • [$ cd build-cocoa-debug]
    • $ ../configure --with-macosx-version-min
    • [$ ../configure –-enable-debug --with-macosx-version-min]
    • $ make
    • $ sudo su
    • make install
    • exit
    • $ cd samples
    • $ make
    • $cd ../demos
    • $make

• OpenCV 4.x

  • Windows (+Visual Studio)
    • Download release files from https://opencv.org/releases/ and extract files
  • Linux (Ubuntu)
    • $ sudo apt update && sudo apt install -y cmake g++ wget unzip
    • $ sudo apt-get install ffmpeg
    • $ git clone https://github.com/opencv/opencv.git
    • $ git clone https://github.com/opencv/opencv_contrib.git
    • $ mkdir -p build && cd build
    • $ cmake -D OPENCV_EXTRA_MODULES_PATH=../opencv_contrib/modules ../opencv
    • $ cmake --build .
    • $ sudo make install
    • $ make
  • Mac
    • $ /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
    • $ brew install cmake
    • $ brew install ffmpeg@4
    • $ brew link ffmpeg@4
    • $ mkdir opencv
    • $ cd opencv
    • $ git clone https://github.com/opencv/opencv.git
    • $ git clone https://github.com/opencv/opencv_contrib.git
    • $ mkdir build
    • $ cd build
    • $ cmake -D BUILD_OPENEXR=ON -D WITH_FFMPEG=ON -D OPENCV_EXTRA_MODULES_PATH=../opencv_contrib/modules ../opencv
    • $ cmake --build .
    • $ sudo make install
    • $ make

• Build

  • MDI (multiple-document interface, default), SDI (single-document interface, define _KHUCV_SDI)
  • Windows (+Visual Studio) I
    • Make new desktop empty project (wxWidgets sample: https://wiki.wxwidgets.org/Microsoft_Visual_C%2B%2B_Guide)
    • Add source/header files
    • Set C++ Language Standard: ISO C++17 Standard (/std:c++17)
    • Add Include and library paths
    • Add opencv_world4xx.lib to Additional Dependencies
    • Build
  • Windows (+Visual Studio) II
    • Run "Developer Command Prompt for VS 20XX"
    • $ cd build-windows
    • $ [MDI] cmake . -A x64 -T host=x64 -DWX_ROOTDIR=[DIR] -DOPENCV_ROOTDIR=[DIR]
      • e.g.) cmake . -A x64 -T host=x64 -DWX_ROOTDIR="G:/Cpp_Libs/wxWidgets/wxWidgets-3.1.6" -DOPENCV_ROOTDIR="G:/Cpp_Libs/OpenCV/OpenCV_64/install"
      • e.g.) cmake . -A x64 -T host=x64 -DWX_ROOTDIR="G:/Cpp_Libs/wxWidgets/wxWidgets-3.1.6" -DOPENCV_ROOTDIR="G:/Cpp_Libs/OpenCV/build"
    • $ [SDI] cmake . -DSDI=True -A x64 -T host=x64 -DWX_ROOTDIR=[DIR] -DOPENCV_ROOTDIR=[DIR]
      • e.g.) cmake . -DSDI=True -A x64 -T host=x64 -DWX_ROOTDIR="G:/Cpp_Libs/wxWidgets/wxWidgets-3.1.6" -DOPENCV_ROOTDIR="G:/Cpp_Libs/OpenCV/OpenCV_64/install"
      • e.g.) cmake . -DSDI=True -A x64 -T host=x64 -DWX_ROOTDIR="G:/Cpp_Libs/wxWidgets/wxWidgets-3.1.6" -DOPENCV_ROOTDIR="G:/Cpp_Libs/OpenCV/build"
    • $ [MDI] msbuild KhuCv.vcxproj /p:configuration=release /p:platform=x64 /p:CharacterSet=Unicode
    • $ [SDI] msbuild KhuCvS.vcxproj /p:configuration=release /p:platform=x64 /p:CharacterSet=Unicode
  • Linux (Ubuntu)
    • $ cd build-linux
    • [SDI] $ cmake . -D SDI=True
    • [MDI] $ cmake . -D SDI=False
    • $ make
  • Mac (+Xcode) I
    • $ cd build-mac
    • [SDI] $ cmake . -G Xcode -D SDI=True
    • [MDI] $ cmake . -G Xcode -D SDI=False
  • Mac (+Xcode) II
    • wxWidgets sample: https://wiki.wxwidgets.org/Creating_Xcode_projects_for_wxWidgets_applications
    • Add source/header files
    • Add Include and library paths
    • Add opencv libraries
    • Build
  • Mac III
    • $ cd build-mac
    • [SDI] $ cmake . -D SDI=True
    • [MDI] $ cmake . -D SDI=False
    • $ make

[Windows-MDI]

[Windows-SDI]

[Linux-Ubuntu-MDI]

[Mac-MDI]

Sequence processing

• Image sequences and video files
• Example
  • Open mp4 file and Run
    

Source code

• Call a method (Run) for image processing

// MainDialog.cpp
void CMainDialog::OnTimer(wxTimerEvent& event) {
	// Timer for sequence processing

	...

	if(bLoaded) {
		auto start = std::chrono::steady_clock::now();
		
		cv::Mat Output;
		
		// Project Run;
		m_Project.Run(CurrentImage, Output, m_pVerboseCheck->GetValue());
    		// m_Porject is a CProject class instance for sequence processing
   		// Since you only modify the CProject class, you can easily show and debug the processing and results
    
   		// Display codes
		...
	}

	...
}

• CProject class

// CProcessing.cpp / h
class CProject
{
	cv::Mat m_PreviousImage;
public:
	CProject();
	void Run(cv::Mat Input, cv::Mat& Output, bool bVerbose);
};

CProject::CProject() {
}

void CProject::Run(cv::Mat Input, cv::Mat& Output, bool bVerbose) {
	cv::Sobel(Input, Output, -1, 1, 1);

	cv::Mat Difference;

	if (!m_PreviousImage.empty()) {

		if (Input.type() == m_PreviousImage.type() &&
			Input.cols == m_PreviousImage.cols &&
			Input.rows == m_PreviousImage.rows) {
			cv::subtract(Input, m_PreviousImage, Difference);

			if (bVerbose) DisplayImage(Difference, 0, Input.rows, false, false);
		}
	}

	m_PreviousImage = Input.clone();
}

• Showing string information (DlgPrintf function) - lower output window

DlgPrintf("%05d: %s, %10.5lfms", m_nProcessingNum, fileName, processingTime);

References

• The Common Objects in Context (COCO) images and Title-based Video Summarization (TVSum)/SumMe video files are used for demonstration.
• https://cocodataset.org/
• http://people.csail.mit.edu/yalesong/tvsum/
• https://gyglim.github.io/me/vsum/index.html
• Intstall wxWidgets: https://www.cs.sjsu.edu/~mak/tutorials/InstallwxWidgets.pdf