Yan
Yan Just a simple man.

编译并打包OpenCV

编译并打包OpenCV

这篇记录如何在arm环境下编译opencv并打包成deb包。

获取源码

  1. Clone源码
1

yan@yan-wyb:~$ git clone https://github.com/opencv/opencv.git
  1. 切换到需要的tag

这里以4.5为例，

1
2

yan@yan-wyb:~$ cd opencv/
yan@yan-wyb:~/opencv$ git checckout -b 4.5.0

可以使用git tag命令查看所有可用的tag

编译

  1. opencv编译是使用cmake命令。没有的请先安装cmake，以ubuntu为例
1

yan@yan-wyb:~$ sudo apt install cmake
  1. 创建编译目录
1
2

yan@yan-wyb:~$ cd opencv
yan@yan-wyb:~/opencv$ mkdir build
  1. 使用cmake编译

这里编译使用一些常见选项，并安装到当前目录

1
2
3
4
5
6
7
8
9
10
11
12
13
14

yan@yan-wyb:~/opencv$ cd build/
yan@yan-wyb:~/opencv/build$ cmake -D CMAKE_BUILD_TYPE=RELEASE \
      -D CMAKE_INSTALL_PREFIX=./install \
      -D BUILD_TESTS=OFF \
      -D BUILD_EXAMPLES=OFF \
      -D INSTALL_PYTHON_EXAMPLES=OFF \
      -D BUILD_opencv_gapi=OFF \
      -D BUILD_opencv_python2=OFF \
      -D BUILD_opencv_python3=ON \
      -D PYTHON3_INCLUDE_DIR=/usr/include/python3.8 \
      -D PYTHON3_EXECUTABLE=/usr/bin/python3.8 \
      -D PYTHON3_LIBRARY=/usr/lib/aarch64-linux-gnu/libpython3.8.so.1.0 \
      -D PYTHON3_NUMPY_INCLUDE_DIRS=/usr/lib/python3.8/dist-packages/numpy/core/include/
yan@yan-wyb:~/opencv/build$ make

等待编译完成，编译完成以后安装到指定的目录，即当前目录下

1

yan@yan-wyb:~/opencv/build$ make install

如要安装到其他目录可在上一步中指定安装位置-D CMAKE_INSTALL_PREFIX=

打包

  1. 创建打包脚本
1
2
3

yan@yan-wyb:~/opencv/build$ cd install
yan@yan-wyb:~/opencv/build/install$ mkdir DEBIAN
yan@yan-wyb:~/opencv/build/install$ vim DEBIAN/control

内容如下

1
2
3
4
5
6
7
8

Package: opencv
Version: 4.5.0
Section: kernel
Architecture: arm64
Maintainer: Yan <yan-wyb@foxmail.com>
Installed-Size: 1
Priority: optional
Description: OpenCV libraries.
  1. 打包
1
2

yan@yan-wyb:~$ cd opencv/build/
yan@yan-wyb:~$ dpkg-deb -b install ./opencv-4.5.0.deb