ubuntu16.04 在cuda9.0环境下编译安装opencv2.4.13.7
ubuntu16.04 在cuda9.0环境下编译安装opencv2.4.13.7
安装步骤:
1.安装cuda9.0：https://blog.csdn.net/zhuangwu116/article/details/81063234
2.安装依赖包:
sudo apt-get install build-essential
sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev
sudo apt-get install cmake git libgtk2.0-dev pkg-config这里还需要安装解码音视频库，安装解码音视频库有两种方法:
（1）一种是编译安装：
（2）另一种是命令安装：
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev如何已经编译安装解码视频库的可以跳过，不需要命令安装，以免冲突。
3.下载代码:
git clone https://github.com/opencv/opencv.git
cd opencv
git checkout -b 2.4.13.7或者你也可以直接下载压缩包 地址
4.修改opencv编译脚本
修改opencv编译的脚本是因为两点:
1.在cuda9里面，NVIDIA把 libnppi.so换成libnppc.so libnppial.so libnppicc.so libnppicom.so libnppidei.so libnppif.so libnppig.so libnppim.so libnppist.so libnppisu.so libnppitc.so libnpps.so
2.cuda9里面不支持‘ compute-20 ’
1.解决第一个问题：
修改opencv/cmake/FindCUDA.cmake 文件，将524行的
unset(CUDA_nppi_LIBRARY CACHE)修改为
unset(CUDA_nppial_LIBRARY CACHE)
unset(CUDA_nppicc_LIBRARY CACHE)
unset(CUDA_nppicom_LIBRARY CACHE)
unset(CUDA_nppidei_LIBRARY CACHE)
unset(CUDA_nppif_LIBRARY CACHE)
unset(CUDA_nppig_LIBRARY CACHE)
unset(CUDA_nppim_LIBRARY CACHE)
unset(CUDA_nppist_LIBRARY CACHE)
unset(CUDA_nppisu_LIBRARY CACHE)
unset(CUDA_nppitc_LIBRARY CACHE)修改opencv/cmake/FindCUDA.cmake 文件，将799行和801行的
find_cuda_helper_libs(nppi)
set(CUDA_npp_LIBRARY"${CUDA_nppc_LIBRARY};${CUDA_nppi_LIBRARY};${CUDA_npps_LIBRARY}")修改为
find_cuda_helper_libs(nppial)
find_cuda_helper_libs(nppicc)
find_cuda_helper_libs(nppicom)
find_cuda_helper_libs(nppidei)
find_cuda_helper_libs(nppif)
find_cuda_helper_libs(nppig)
find_cuda_helper_libs(nppim)
find_cuda_helper_libs(nppist)
find_cuda_helper_libs(nppisu)
find_cuda_helper_libs(nppitc)
set(CUDA_npp_LIBRARY "${CUDA_nppc_LIBRARY};${CUDA_nppial_LIBRARY};${CUDA_nppicc_LIBRARY};${CUDA_nppicom_LIBRARY};${CUDA_nppidei_LIBRARY};${CUDA_nppif_LIBRARY};${CUDA_nppig_LIBRARY};${CUDA_nppim_LIBRARY};${CUDA_nppist_LIBRARY};${CUDA_nppisu_LIBRARY};${CUDA_nppitc_LIBRARY};${CUDA_npps_LIBRARY}")2.解决第二个问题方案
更改 OpenCVDetectCUDA.cmake 文件，把有关 ‘ compute-20 ’ 的修改掉。
将74行中的
73 if(CUDA_GENERATION STREQUAL "Fermi")
74 set(__cuda_arch_bin "2.0 2.1(2.0)")
75 elseif(CUDA_GENERATION STREQUAL "Kepler")
76 if(${CUDA_VERSION} VERSION_LESS "5.0")
77 set(__cuda_arch_bin "3.0")
78 else()
79 set(__cuda_arch_bin "3.0 3.5")
80 endif()修改成
73 if(CUDA_GENERATION STREQUAL "Fermi")
74 set(__cuda_arch_bin "3.0 3.5")
75 elseif(CUDA_GENERATION STREQUAL "Kepler")
76 if(${CUDA_VERSION} VERSION_LESS "5.0")
77 set(__cuda_arch_bin "3.0")
78 else()
79 set(__cuda_arch_bin "3.0 3.5")
80 endif()将107行
104 if(${CUDA_VERSION} VERSION_LESS "5.0")
105 set(__cuda_arch_bin "1.1 1.2 1.3 2.0 2.1(2.0) 3.0")
106 elseif(${CUDA_VERSION} VERSION_GREATER "6.5")
107 set(__cuda_arch_bin "2.0 2.1(2.0) 3.0 3.5")
108 else()
109 set(__cuda_arch_bin "1.1 1.2 1.3 2.0 2.1(2.0) 3.0 3.5")
110 endif()
111 set(__cuda_arch_ptx "3.0")
112 endif()修改成
104 if(${CUDA_VERSION} VERSION_LESS "5.0")
105 set(__cuda_arch_bin "1.1 1.2 1.3 2.0 2.1(2.0) 3.0")
106 elseif(${CUDA_VERSION} VERSION_GREATER "6.5")
107 set(__cuda_arch_bin "3.0 3.5")
108 else()
109 set(__cuda_arch_bin "1.1 1.2 1.3 2.0 2.1(2.0) 3.0 3.5")
110 endif()
111 set(__cuda_arch_ptx "3.0")
112 endif()5.编译安装OpenCV
mkdir release
cd release
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..
make -j7 #此处看您的cpu内核数，本人机器8核
sudo make install6.测试OpenCV
$ python
>>import cv2
>>print cv2.__version__参考：https://docs.opencv.org/2.4.13.7/doc/tutorials/introduction/linux_install/linux_install.html#linux-installation
参考：http://wgshun.top/2018/06/ubuntu-install-opencv/