OpenCV使用指南（v2.1）

2.1 下载代码

    git clone --branch 4.13.0 --depth 1 https://github.com/opencv/opencv.git
    git clone --branch 4.13.0 --depth 1 https://github.com/opencv/opencv_contrib.git
    git clone --branch 4.13.0 --depth 1 https://github.com/opencv/opencv_extra.git
    
    cd opencv
    ln -sf ../opencv_contrib opencv_contrib

2.2 准备脚本

编译前请确认已经安装PPU SDK并执行了envsetup.sh，配置好SDK的环境。

编译删除了一些模块，尽可能多的打开了cuda加速模块，用户可以根据自己的实际需要做增删修改：

mkdir build
cd build

cmake_option=""

# 根据需要配置SKIP modules，这里的skip只是示例，用户可以取消
SKIP_LIST=(xfeatures2d matlab xobjdetect xphoto wechat_qrcode)  
SKIP_DEFINES=''
for m in ${SKIP_LIST[@]}; do
    SKIP_DEFINES+=" -DBUILD_opencv_${m}=OFF"
done
cmake_option+=" ${SKIP_DEFINES}"

# 根据需要增加需要的module，这里打开了尽可能多的cuda加速modules
ENABLE_LIST=(legacy ximgproc cudev cudastereo cudaarithm cudaimgproc cudacodec cudafilters cudawarping cudaoptflow cudabgsegm cudafeatures2d cudaobjdetect cudalegacy)
ENABLE_DEFINES=''
for m in ${ENABLE_LIST[@]}; do
    ENABLE_DEFINES+=" -DBUILD_opencv_${m}=ON"
done
cmake_option+=" -DWITH_CUDA=ON -DWITH_CUDNN=ON -DWITH_CUBLAS=ON -DWITH_CUFFT=ON -DOPENCV_DNN_CUDA=ON -DCUDA_ARCH_BIN=80 -DWITH_NVCUVID=ON"
cmake_option+=" -DOPENCV_ENABLE_NONFREE=ON"
cmake_option+=" ${ENABLE_DEFINES}"

# 关掉了一些不是必须的选项，用户可以取消
cmake_option+=" -DWITH_OPENCL=OFF -DOPENCV_DNN_OPENCL=OFF"
cmake_option+=" -DWITH_ANDROID_MEDIANDK=OFF -DWITH_GTK=OFF -DWITH_IPP=OFF -DBUILD_JAVA=OFF"
cmake_option+=" -DWITH_V4L=OFF -DWITH_GSTREAMER=OFF -DWITH_1394=OFF"

# 指定opencv contrib位置，这里放在opencv目录下一级
cmake_option+=" -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib/modules/"

# 这里指定编译安装位置在源码上一级目录；如果不指定，make install会安装到/usr/local目录
cmake_option+="-DOPENCV_GENERATE_PKGCONFIG -DBUILD_SHARED_LIBS=ON -DCMAKE_INSTALL_PREFIX=../"

2.3 配置FFmpeg支持

如果系统没有安装FFmpeg，在opencv configure时，仅仅打开-DWITH_FFMPEG=ON，编译是无法找到ffmpeg的，最终编译的libopencv_videoio.so不会包含对ffmpeg的支持。

有关FFmpeg的编译和使用，可以参考另一篇指南文档 “Video FFmpeg使用指南”。

假设你的FFmpeg安装目录是 FFMPEG_PATH，在build_opencv.sh中补充如下脚本：

export PKG_CONFIG_LIBDIR=$FFMPEG_PATH/lib/pkgconfig:$PKG_CONFIG_LIBDIR
cmake_option+=" -DWITH_FFMPEG=ON -DOPENCV_FFMPEG_USE_FIND_PACKAGE=ON -DOPENCV_FFMPEG_SKIP_BUILD_CHECK=ON -DFFMPEG_DIR=./"

配置成功，在编译的时候可以看到（Yes后面显示的是本地的FFmpeg版本）：

--   Video I/O:
--     FFMPEG:                      YES (find_package)
--       avcodec:                   YES (60.31.102)
--       avformat:                  YES (60.16.100)
--       avutil:                    YES (58.29.100)
--       swscale:                   YES (7.5.100)
--       avresample:                NO

请把ffmpeg目录指向自编译的支持硬件加速的ffmpeg安装目录。

2.4 配置testcase和samples

增加如下option可以编译并安装opencv自带的testcases和testdata，此项不是必须的：

cmake_option+=" -DBUILD_PERF_TESTS=ON -DBUILD_TESTS=ON -DINSTALL_TESTS=ON -DOPENCV_TEST_DATA_PATH=../opencv_extra/testdata"

这里opencv_extra是test data所在目录，前面的脚本已经从github下载了。

2.5 编译

cmake $cmake_option ..
make -j && make install

2.6 运行示例

进入安装目录下的bin目录：

cd ../bin
export LD_LIBRARY_PATH=$FFMPEG_PATH/lib:../lib:$LD_LIBRARY_PATH  # 把ffmpeg和opencv lib目录加入LD_LIBRARY_PATH
export OPENCV_TEST_DATA_PATH=../share/opencv4/testdata  # testdata是git clone https://github.com/opencv/opencv_extra.git得到的目录

可以执行相关native测试程序。

3.1 下载代码

git clone --branch 90 --depth 1 https://github.com/opencv/opencv-python.git

因为上面我们编译的是opencv 4.13，所以这里需要下载对应的opencv-python版本。

3.2 修改setup.py

我们可以直接使用 #2.1 已经下载好的opencv和opencv_contrib代码，不需要重新下载

cd opencv-python
rm opencv opencv_contrib opencv_extra multibuild -r
# 假设opencv-python, opencv, opencv_contrib都在同一个目录
ln -sf ../opencv opencv
ln -sf ../opencv_contrib opencv_contrib
ln -sf ../opencv_extra opencv_extra
git clone https://github.com/multi-build/multibuild.git

修改setup.py，去掉对submodule的git update，更改编译options（修改项跟上面#2.2差不多，可以根据需要自行增删）：

diff --git a/setup.py b/setup.py
index e25e51b..cb8da9b 100755
--- a/setup.py
+++ b/setup.py
@@ -187,6 +187,49 @@ def main():
             "-DBUILD_DOCS=OFF",
             "-DPYTHON3_LIMITED_API=ON",
             "-DBUILD_OPENEXR=ON",
+            "-DWITH_OPENCL=OFF",
+            "-DOPENCV_DNN_OPENCL=OFF",
+            "-DWITH_FFMPEG=ON",
+            "-DOPENCV_FFMPEG_USE_FIND_PACKAGE=ON",
+            "-DOPENCV_FFMPEG_SKIP_BUILD_CHECK=ON",
+            "-DWITH_CUDA=ON",
+            "-DWITH_CUDNN=ON",
+            "-DWITH_CUBLAS=ON",
+            "-DWITH_CUFFT=ON",
+            "-DOPENCV_DNN_CUDA=ON",
+            "-DCUDA_ARCH_BIN=80",
+            "-DWITH_NVCUVID=ON",
+            "-DWITH_V4L=OFF",
+            "-DWITH_GSTREAMER=OFF",
+            "-DWITH_1394=OFF",
+            "-DWITH_ANDROID_MEDIANDK=OFF",
+            "-DWITH_GTK=OFF",
+            "-DWITH_IPP=OFF",
+            "-DBUILD_JAVA=OFF",
+            "-DWITH_VTK=OFF",
+            "-DENABLE_FLAKE8=OFF",
+            "-DENABLE_PYLINT=OFF",
+            "-DBUILD_opencv_legacy=ON",
+            "-DOPENCV_ENABLE_NONFREE=ON",
+            "-DBUILD_opencv_xfeatures2d=OFF",
+            "-DBUILD_opencv_matlab=OFF",
+            "-DBUILD_opencv_xobjdetect=OFF",
+            "-DBUILD_opencv_xphoto=OFF",
+            "-DBUILD_opencv_wechat_qrcode=OFF",
+            "-DBUILD_opencv_ximgproc=ON",
+            "-DBUILD_opencv_cudev=ON",
+            "-DBUILD_opencv_cudastereo=ON",
+            "-DBUILD_opencv_legacy=ON",
+            "-DOPENCV_ENABLE_NONFREE=ON",
+            "-DBUILD_opencv_xfeatures2d=OFF",
+            "-DBUILD_opencv_matlab=OFF",
+            "-DBUILD_opencv_xobjdetect=OFF",
+            "-DBUILD_opencv_xphoto=OFF",
+            "-DBUILD_opencv_wechat_qrcode=OFF",
+            "-DBUILD_opencv_ximgproc=ON",
+            "-DBUILD_opencv_cudev=ON",
+            "-DBUILD_opencv_cudastereo=ON",
         ]
         + (
             # CMake flags for windows/arm64 build

3.3 编译

依赖skbuild，可以 pip install scikit-build

export ENABLE_CONTRIB=1
python3 setup.py bdist_wheel

3.4 安装

编译生成的opencv_contrib_python-4.13.0.90-cp37-cp37m-linux_x86_64.whl在dist目录下

pip install dist/opencv_contrib_python-4.13.0.90-cp37-cp37m-linux_x86_64.whl

此时可以进入python交互页面，输入 import cv2看看能否成功。