Building OpenCV from source

Install build dependencies

Update the system

sudo apt-get update && sudo apt-get upgrade

Install development tools and CMake

sudo apt-get install build-essential cmake unzip pkg-config

Image and video codecs

The original notes suggest installing all of these:

JPEG, PNG and TIFF: sudo apt-get install libjpeg-dev libpng-dev libtiff-dev
Video capture, including cameras: sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev
Video files and H.264: sudo apt-get install libxvidcore-dev libx264-dev
Additional codecs: sudo apt-get install libdc1394-22-dev libtheora-dev libvorbis-dev yasm libopencore-amrnb-dev libopencore-amrwb-dev libxine2-dev

GUI tools

Install GTK if using graphical output such as an OpenCV window:

sudo apt-get install libgtk-3-dev libgtkglext1-dev

For GTK-related support:

sudo apt-get install libcanberra-gtk*

For Qt:

sudo apt-get install qt5-default

For VT:

sudo apt-get install libvtk6-dev

Numerical optimisation tools

The original notes add OpenCV’s numerical libraries for numerical processing, while noting a possible performance trade-off:

sudo apt-get install libatlas-base-dev gfortran

Optimisation and parallel processing

sudo apt-get install -y libtbb-dev libeigen3-dev

Python bindings

For Python 3:

sudo apt-get install python3-dev

For Python 2:

sudo apt-get install python-dev

Java bindings

sudo apt-get install -y ant default-jdk

Download OpenCV sources

Choose a release on the GitHub releases page, then download with wget:

cd ~
wget -O opencv.zip https://github.com/opencv/opencv/archive/4.x.y.zip
wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.x.y.zip

Extract and rename the directories:

unzip opencv.zip
unzip opencv_contrib.zip

mv opencv-4.0.0 opencv
mv opencv_contrib-4.0.0 opencv_contrib

Build environment

Install into the main system

The original notes offer this approach if not using virtualenv.

Install NumPy system-wide:

sudo pip3 install numpy

Or for the user:

pip3 install numpy --user

Use virtualenv

A virtual environment isolates Python packages from the main system.

Install virtualenv:

sudo pip install virtualenv virtualenvwrapper
sudo rm -rf ~/get-pip.py ~/.cache/pip

Add it to the shell environment through ~/.profile:

echo -e "# virtualenv and virtualenvwrapper \
export WORKON_HOME=$HOME/.virtualenvs  \
export VIRTUALENVWRAPPER_PYTHON=/usr/bin/python3 \
source /usr/local/bin/virtualenvwrapper.sh" >> ~/.profile

Reload the shell configuration:

source ~/.profile

Create a Python environment:

mkvirtualenv cv -p python3

Enter it:

workon cv

In later sessions, use workon cv to enter the environment named cv.

Install NumPy:

pip3 install numpy

Build OpenCV

Configure

OpenCV is predominantly C++ and uses CMake. Keep build output separate from the source by creating a build directory inside the OpenCV checkout:

cd ~/opencv
mkdir build
cd build

Configure with CMake:

cmake -D CMAKE_BUILD_TYPE=RELEASE \
    -D CMAKE_INSTALL_PREFIX=/usr/local \
    -D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib/modules \
    -D ENABLE_NEON=ON \
    -D ENABLE_VFPV3=ON \
    -D WITH_OPENGL=ON \
    -D WITH_QT=ON \
    -D FORCE_VTK=ON \
    -D WITH_LIBV4L=ON \
    -D WITH_TBB=ON \
    -D WITH_GDAL=ON \
    -D WITH_FFMPEG=0 \
    -D WITH_XINE=ON \
    -D BUILD_TESTS=ON \
    -D OPENCV_ENABLE_NONFREE=ON \
    -D ENABLE_PRECOMPILED_HEADERS=OFF \
    -D INSTALL_PYTHON_EXAMPLES=ON \
    -D BUILD_EXAMPLES=ON ..

Compile

Start Make:

make

For a multicore CPU, use a parallel build:

make -j{çekirdek sayınızın bir eksiği}

For example, on four cores:

make -j3

Install

sudo make install
sudo ldconfig

When run in the virtual environment, this installs the Python binding there. The C++ libraries and shared binaries are installed in the main system.