OpenCV3.2.0+OpenCV_contrib3.2.0+cuda8.0安装相关

cuda8.0 + cudnn + NVIDIA-i384安装

cuda8.0 + cudnn + NVIDIA-i384安装
（可以在系统中下载NVIDIA驱动，不用关闭X-window显示服务）

下载

在github上下载OpenCV3.2.0+OpenCV_contrib3.2.0

安装opencv依赖库

$ sudo apt-get install build-essential
$ sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev
$ sudo apt-get install libgtk2.0-dev pkg-config python-dev python-numpy libdc1394-22 libdc1394-22-dev libjpeg-dev libpng12-dev libtiff5-dev libjasper-dev libavcodec-dev libavformat-dev libswscale-dev libxine2-dev libgstreamer0.10-dev libgstreamer-plugins-base0.10-dev libv4l-dev libtbb-dev libqt4-dev libfaac-dev libmp3lame-dev libopencore-amrnb-dev 
$ sudo apt-get install libgtk-3-dev
$ sudo apt-get install libgstreamer0.10-dev
$ sudo apt-get install libgstreamer-plugins-base0.10-dev
$ sudo apt-get install libgstreamer1.0-dev 
$ sudo apt-get install libgstreamer-plugins-base1.0-dev 
$ sudo apt-get install libgstreamer1.0-dev 
$ sudo apt-get install libgstreamer-plugins-base0.10-dev
$ sudo apt-get install libgstreamer0.10-dev 
$ sudo apt-get install libgstreamer-plugins-base1.0-dev
$ sudo apt-get install libavresample-dev
$ sudo apt-get install libgphoto2-dev
$ sudo apt-get install libavresample-dev
$ sudo apt-get install libgphoto2-dev
$ sudo apt-get install libopenblas-dev
$ sudo apt-get install doxygen
$ sudo apt-get install liblapacke-dev checkinstall

解压opencv-3.2.0、opencv_contrib-3.2.0.tar.gz文件

$ unzip opencv-3.2.0.zip
$ tar -zxvf opencv_contrib-3.2.0.tar.gz -C ./opencv-3.2.0/

编译安装opencv（无cuda版本）

$ cd opencv-3.2.0/
$ mkdir build
$ cd build
$ cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local -D OPENCV_EXTRA_MODULES_PATH=/home/ai/downloads/opencv-3.2.0/opencv_contrib-3.2.0/modules/ -D WITH_TBB=ON -D BUILD_SHARED_LIBS=OFF -D WITH_OPENMP=ON -D ENABLE_PRECOMPILED_HEADERS=OFF ..
$ sudo make -j8  
$ sudo make install

编译安装opencv（带cuda版本）

$ cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/home/jiang/6_lib/install/opencv3.2.0 -D OPENCV_EXTRA_MODULES_PATH=/home/jiang/6_lib/opencv_contrib-3.2.0/modules -D WITH_CUDA=ON -D WITH_CUBLAS=1 -D CUDA_ARCH_BIN="6.1" -D CUDA_ARCH_PTX="6.1" -D INSTALL_C_EXAMPLES=OFF -D INSTALL_PYTHON_EXAMPLES=ON -D WITH_OPENGL=ON -D WITH_V4L=ON WITH_NVCUVID=ON ..
或者
$ cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_CXX_FLAGS="-std=c++11" -D CUDA_NVCC_FLAGS="-std=c++11 --expt-relaxed-constexpr" -D CMAKE_INSTALL_PREFIX=/home/jiang/6_lib/install/opencv3.3.0 -D OPENCV_EXTRA_MODULES_PATH=/home/jiang/6_lib/opencv-3.3.0/opencv_contrib-3.3.0/modules -D WITH_CUDA=ON -D ENABLE_FAST_MATH=1 -D CUDA_FAST_MATH=1 -D WITH_CUBLAS=1 -D CUDA_ARCH_BIN="6.1" -D CUDA_ARCH_PTX="6.1" -D INSTALL_C_EXAMPLES=OFF -D INSTALL_PYTHON_EXAMPLES=ON -D WITH_OPENGL=ON -D WITH_V4L=ON WITH_NVCUVID=ON ..
$ make -j8
$ make install

注意

1.CMAKE_INSTALL_PREFIX默认目录为/usr/local，可自行选择，只需要在更新环境变量的时候对应更改就行。
2.CUDA_ARCH_BIN和CUDA_ARCH_PTX这里是指的显卡计算能力，GeForce GTX 1050～1080Ti 的计算能力为6.1，其他型号可以自行上NVIDIA官网查询对应版本：
https://developer.nvidia.com/cuda-gpus
3.cmake的时候，有些contrib模块编译时需要的文件可能下载不下来，具体信息可以到build目录下CMakeDownloadLog.txt查看，去对应网址下载好文件放到opencv-3.4.1/.cache目录下（注意这是个隐藏文件）的对应文件夹里，并重命名（加上前面的一串字符，注意要一一对应），再重新cmake。下载的缺少的文件（data，ippicv，tiny_dnn，dnn_face_detector，xfeatures2d已经重命名好）。

配置环境

$ sudo gedit /etc/ld.so.conf.d/opencv.conf
# 在最后一行添加 /usr/local/lib
$ sudo gedit /etc/bash.bashrc
# 在最后一行中添加：
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/opencv320/lib/pkgconfig  
export PKG_CONFIG_PATH
$ sudo ldconfig
$ sudo cp /usr/local/lib/pkgconfig/opencv.pc /usr/lib/pkgconfig

测试安装成功

$ pkg-config --modversion opencv

参考文献

[1] https://www.cnblogs.com/Jessey-Ge/p/10872331.html
[2] https://www.jianshu.com/p/1baf9848b7d7