Package for building OpenCV 4.0.0-alpha including Python 3 bindings from the official sources.
Unofficial OpenCV builder for Python.
This package aims at building OpenCV with Python bindings from the official sources. It provides a simple command line interface for starting the process of downloading the official sources, configuring the build dependencies, compiling and installing the resulting CV2 shared object within a virtual environment.
In contrast to opencv-python it will not provide any wheels and therefore the installation / build process will be by far slower (depending on the actual system performance).
Depending on the usage and system dependencies, the on-the-fly build output can contain video and GUI functionality and the contrib package.
- builds against Python >= 3.4
- runs and builds inside virtual environment
- includes video support
- supports OpenCV check (import, build information)
- compiles with many flags enabled (which???)
- Debian Jessie, Stretch
- Ubuntu 16.04
- LinuxMint 18.2
- Raspbian 9 (Stretch)
Supported Python runtimes
- CPython 3.4, 3.5, 3.6, 3.7
- PyPy (not yet)
Note: PyPy (6.0.0) is not supported yet due to missing path variables in module
- Use a Python's virtual environment
or even better add the package via poetry:
poetry add cvbuilder
- Follow instructions below
After installing the package via
poetry, you can manually invoke
Install system dependencies
cvbuilder system --enable-gui --enable-video
Download, configure, build, install
do-it-all command is:
or if you already downloaded the source zip files in the default temporary directory or need to rerun the process in a clean way:
cvbuilder build --clean
If you need to run the individual steps (i.e. for debugging) the following commands are provided.
Generate make config
cvbuilder configure [--tmpdir XYZ]
You can run the check command:
in order to check whether the build process and installation worked.
Now everything should be up and running and you should be able to work with OpenCV:
- Start a python REPL:
pipenv run python
- Import the CV package:
- Read OpenCV documentation