OpenCV2Qml

This library allow to use OpenCV's video output on a qml application. This library has been developed and tested on Mac on Mojave OS.

Installation

Dependencies

To use this library start by installing OpenCV 3.4.5 .

MacOS

On MacOS install the lib with homebrew :

/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

With brew install QT and OpenCV :

brew install opencv@3
brew install qt

Make sure the .pro has the right informations. The paths should match the path to your opencv.

INCLUDEPATH += /usr/local/Cellar/opencv/3.4.5/include
INCLUDEPATH += /usr/local/Cellar/opencv/3.4.5/lib

LIBS += \
    -L/usr/local/Cellar/opencv/3.4.5/lib \
    -lopencv_bgsegm -lopencv_shape -lopencv_video \
    -lopencv_highgui -lopencv_videoio -lopencv_flann \
    -lopencv_xobjdetect -lopencv_imgcodecs -lopencv_objdetect \
    -lopencv_xphoto -lopencv_imgproc -lopencv_core \
    -lopencv_face

Configuration

Before generating the library you need to do some configuration to make the lib suit your need. Find configopencv.h in includes. This files contains some elements altering the way the program works, like the number of FPS amongst other things.

Usage

Once the lib generated you can use it like any other lib. To use it from another QT project do so :

PRE_TARGETDEPS  += /PATH/TO/libOpenCVtoQml.1.0.0.dylib
LIBS            += -L/PATH/TO/DIRECTORY/ROOT/OF/OPENCVTOQML -lOpenCVtoQml.1.0.0
INCLUDEPATH     += /$$PWD/PATH/TO/LIB/INCLUDES/
DEPENDPATH      += /$$PWD/PATH/TO/DIRECTORY/ROOT/OF/OPENCVTOQML/libOpenCVtoQml.1.0.0.dylib

Example :

For a repository organized like that :

├── ShaderFun.pro
├── ShaderFun.pro.user
├── ShaderFun.pro.user.4.8-pre1
├── ShadersList.qml
├── includes
│   ├── Config.h
│   ├── FileReader.h
│   └── ShaderManager.h
├── lib
│   └── cv2qml
│       ├── libOpenCVtoQml\ 2.dylib -> libOpenCVtoQml.1.0.0.dylib
│       ├── libOpenCVtoQml.1.0.0.dylib
│       ├── libOpenCVtoQml.1.0.dylib -> libOpenCVtoQml.1.0.0.dylib
│       ├── libOpenCVtoQml.1.dylib -> libOpenCVtoQml.1.0.0.dylib
│       ├── libOpenCVtoQml.a
│       ├── libOpenCVtoQml.dylib -> libOpenCVtoQml.1.0.0.dylib
│       └── release
│           ├── Buffer.h
│           ├── CaptureThread.h
│           ├── FaceDetection.h
│           ├── MatToQImage.h
│           ├── OpenCVtoQml.h
│           ├── OpenCVtoQml_global.h
│           ├── SharedImageBuffer.h
│           ├── Structures.h
│           └── configopencv.h
├── main.cpp
├── qml
│   ├── Effect.qml
│   └── main.qml
├── qml.qrc
├── shaders
│   ├── billboard.fsh
│   ├── blueseeker.fsh
│   ├── displacement.fsh
│   ├── glow.fsh
│   ├── outline.fsh
│   ├── redseeker.fsh
│   ├── split.fsh
│   ├── thermalCam.fsh
│   ├── warhol.fsh
│   └── yellowseeker.fsh
└── src
    ├── FileReader.cpp
    └── ShaderManager.cpp

The .pro file would have this configuration :

PRE_TARGETDEPS  += /$$PWD/lib/cv2qml/libOpenCVtoQml.1.0.0.dylib
LIBS            += -L/$$PWD/lib/cv2qml -lOpenCVtoQml.1.0.0
INCLUDEPATH     += /$$PWD/lib/cv2qml/release/
DEPENDPATH      += /$$PWD/lib/cv2qml/libOpenCVtoQml.1.0.0.dylib

Don't forget to add openCV to the new project the same way you did for the library. Then simply #include <OpenCVtoQml.h> in your Cpp to start using it.

How to use

Video Output

What this lib does is defining a new QML element containing OpenCV's output. To use this element start in the main by registering OpenCVtoQml :

qmlRegisterType<OpenCVtoQml>("com.company.name", 1, 0, "OCVRenderer");

Then declare it in your QML :

OCVRenderer {
    id: mediaplayer
    anchors.fill: parent 
    fillMode: OCVRenderer.PreserveAspectCrop
    mode: "camera" // You can choose here between camera and video.
    cameraId: 0 // Id of the camera by default on MacOS the last camera plugged is the 0
    cameraWidth: parent.width    // \
                                 //   > Resolution of the camera's recording
    cameraHeight: parent.height  // /
    videoSource: "camera" // If you chose the mode camera here goes the video's path
    loop: true
    fps: 30
    Component.onCompleted: { 
        mediaplayer.start();
    }
    faceDetectionFlag: true //Should the program run face detection
    haarCascade: "/usr/local/Cellar/opencv/3.4.5/share/OpenCV/haarcascades/haarcascade_frontalface_alt2.xml" //Which haar cascade is gonna be used for face detection
}

Face Detection

The Face detection integration is still a Work In Progress

In the qml OCVRenderer object you can set faceDetectionFlag to true to have the facedetection running. You need to set the haarcascade the file helping with it haarCascade. Here is the list of different options you may use.

.
├── haarcascade_eye.xml
├── haarcascade_eye_tree_eyeglasses.xml
├── haarcascade_frontalcatface.xml
├── haarcascade_frontalcatface_extended.xml
├── haarcascade_frontalface_alt.xml
├── haarcascade_frontalface_alt2.xml
├── haarcascade_frontalface_alt_tree.xml
├── haarcascade_frontalface_default.xml
├── haarcascade_fullbody.xml
├── haarcascade_lefteye_2splits.xml
├── haarcascade_licence_plate_rus_16stages.xml
├── haarcascade_lowerbody.xml
├── haarcascade_profileface.xml
├── haarcascade_righteye_2splits.xml
├── haarcascade_russian_plate_number.xml
├── haarcascade_smile.xml
└── haarcascade_upperbody.xml

Then the OCVRenderer will have a std::vector<cv::Rect> faces property exposed containing the coordinates of the angles around the faces found

Upcoming improvement

Finish Face Detection.
More flexibility. (Right now, the video output is sent as is, with no alteration to it)

Faz95210/Cv2Qml