Introduction

Here I’m going to try to give a “simple” guide to create a virtualenv containing opencv3 with video enabled.

First download a bunch of library

  • First make sure that you are up to date with your repositories:

    sudo apt-get update

  • Download the build tools

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

  • Install some video libraries:

    sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev libgtk2.0-dev

  • (Not mandatory) You can install some libraries that should optimize OpenCV (unfortunately it looks like magic to me):

    sudo apt-get install libatlas-base-dev gfortran

  • Ensure that you have python 2 installed:

    python --version

This should yield:

 Python 2.7.13 #or any other like Python 2.7.x

Now let’s configure the virtualenv

  • Ensure that you have virtualenv helpers installed (plenty of resources on the net like this one)

  • Create a Python2 virtualenv (I will name it very originally “opencv3”):

    mkvirtualenv opencv3 --python=/usr/bin/python2.7

  • Switch to it if it was not done by default, make sure that your PS1 looks like that in your shell, particularly the name of the virtualenv should appear:

    (opencv3) [rest of the PS1]

Mine for instance is: (opencv3) ┌─[ado@parrot]─[~]

  • Install the python build tools:

    sudo apt-get install python-dev

  • Install numpy:

    pip install numpy

Now let’s compile and install opencv3

  • Clone the git repository:

    git clone https://github.com/Itseez/opencv.git

  • Go into the directory:

    cd opencv

  • Small point, perhaps not valid today, but last time I had an issue with the compilation and I had to use the 3.2.0 branch of opencv (it might be fixed or not, but the below branch works at least of today):

    git checkout 3.2.0

  • Create a build directory and switch to it:

    mkdir build && cd build

  • Prepare the build:

    cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..

  • Check that in the report there is a line like:

    FFMPEG: YES

  • Still in the build directory, start the compilation:

    make -j2 #the 2 means that the build will use 2 jobs in parallel, you can increase it depending on your system, as a rule of thumb you should use -jX, X being the number of cores you have

  • Go drink a few coffees (you deserve it)

  • If no errors appeared, install the compiled program:

    sudo make install

  • Add the new dynamic libraries installed:

    sudo ldconfig

The final touch

  • Now link the cv2 library to your virtualenv, normally the following should be ok if you followed a standard install of the components (path of python2 site packages, and creation of the “opencv3” virtualenv):

    ln -s /usr/local/lib/python2.7/site-packages/cv2.so ~/.virtualenvs/opencv3/lib/python2.7/site-packages/cv2.so

  • Now try the below from a python shell, and while having the opencv3 virtualenv active:

     import cv2
     cv2.__version__ #should yield 3.2.0
     video_capture = cv2.VideoCapture(0)
     ret, frame = video_capture.read()
     print ret  # should yield True

If it works congratulations, you have a working virtualenv with opencv3 and video compatible.

P.S: An article that helped me a lot to successfully make this installation