Building OpenCV, especially when building for a different architecture, is a very painful and long process. This repository contains a prebuilt version of OpenCV 3.3.0 that runs under Alpine Linux on the ARMv7 architecture (e.g. a Raspberry Pi 3). It also has a Dockerfile to get started quickly - or to build your own work on.
Note that OpenCV is compiled for Alpine 3.6, and against Python 3.6. Other versions will not work.
- Copy file
opencv-prebuilt/cv2.soto/usr/lib/python3.6/site-packages/cv2.so - Copy directory
opencv-prebuilt/include-opencvto/usr/local/include/opencv - Copy directory
opencv-prebuilt/include-opencv2to/usr/local/include/opencv2 - Copy the content of
opencv-prebuilt/local-libinto/usr/local/lib
After this you can install numpy and other libraries via pip.
-
Install Docker.
-
Build the container:
$ docker build -t alpine-opencv-demo .Note: This still builds Pillow and numpy, which still takes a bit of time.
-
Open a shell into the container:
$ docker run -i -t alpine-opencv-demo /bin/bash -
Run the demo:
$ cd home $ python corner-demo.py # inspect 'chessboard-corners.jpg' for the output image
demo/corner-demo.pyfrom the OpenCV documentation, licensed under the 3-clause BSD License (LICENSE).demo/chessboard.jpgfrom Staxringold via Wikipedia - CC-BY-SA 3.0- OpenCV is licensed under the 3-clause BSD License (LICENSE).
- Rest of the repository is Apache 2.0 licensed.