1200字范文 > Ubuntu16.04 安装 CUDA CUDNN OpenCV 并用 Anaconda 配置 Tensorflow 和 Ca

Ubuntu16.04 安装 CUDA CUDNN OpenCV 并用 Anaconda 配置 Tensorflow 和 Ca

时间：2022-02-22 15:14:14

Ubuntu16.04 安装 CUDA、CUDNN、OpenCV 和 Caffe 详细过程（基于Python2，没有anaconda2和3，因为基于anaconda2和3在安装caffe时会报错，很难解决）

详细安装教程：/zhuiqiuzhuoyue583/article/details/88756053

下面方法不好，会出现很多问题，不容易解决。

1.安装驱动

找到“System Settings”，在找到“Details”。

查看驱动，我这里安装后显示的。

安装NVIDIA驱动：

在“System Settings”找到“Software & Updates ”

再找到“Additional Drivers”，系统会自动搜索可用的 NVIDIA驱动，选择 “NVIDIA”驱动即可。

再去“System Settings”，在找到“Details”，查看驱动：

2.安装相关依赖项

安装后续步骤或环境必需的依赖包，依次输入以下命令:

sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler sudo apt-get install --no-install-recommends libboost-all-dev sudo apt-get install libopenblas-dev liblapack-dev libatlas-base-dev sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev sudo apt-get install git cmake build-essential

有一定几率安装失败而导致后续步骤出现问题，所以要确保以上依赖包都已安装成功，验证方法就是重新运行安装命令，如验证 git cmake build-essential是否安装成功共则再次运行以下命令：

sudo apt-get install git cmake build-essential

界面提示如下则说明已成功安装依赖包，否则继续安装直到安装成功。

yhao@yhao-X550VB:~$ sudo apt-get install git cmake build-essential 正在读取软件包列表... 完成正在分析软件包的依赖关系树正在读取状态信息... 完成 build-essential 已经是最新版 (12.1ubuntu2)。 cmake 已经是最新版 (3.5.1-1ubuntu3)。 git 已经是最新版 (1:2.7.4-0ubuntu1.1)。下列软件包是自动安装的并且现在不需要了： lib32gcc1 libc6-i386 使用'sudo apt autoremove'来卸载它(它们)。升级了 0 个软件包，新安装了 0 个软件包，要卸载 0 个软件包，有 94 个软件包未被升级。

3.安装CUDA

CUDA是NVIDIA的编程语言平台，想使用GPU就必须要使用cuda。

（1）下载CUDA8.0

首先在官网上(/cuda-downloads)下载CUDA，选择自己合适的版本。

该链接界面只显示最新版本。若需要下载以前的版本，可在页面下方点击Legacy Releases，选择自己需要的其他版本，这里安装的是cuda8.0。

这里写图片描述

在Windows中使用迅雷下载，然后再Ubuntu中使用，这样比较快。

图2.CUDA下载

（2）安装CUDA

下载完成后执行以下命令：

sudo chmod 777 cuda_8.0.44_linux.run sudo ./cuda_8.0.44_linux.run

（注意：执行后会先出现一个声明，需要阅读到100%，同意声明后才会开始安装。）

安装时首先会有一系列提示让你确认，但是注意，有个让你选择是否安装nvidia361驱动时，一定要选择否：

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 361.62?

因为前面我们已经安装了更加新的nvidia384，所以这里不要选择安装。其余的都直接默认或者选择是即可。

可能出现的错误：

安装cuda时可能有下面的信息：

Installing the CUDA Toolkit in /usr/local/cuda-8.0 … Missing recommended library: libGLU.so Missing recommended library: libX11.so Missing recommended library: libXi.so Missing recommended library: libXmu.so

原因是缺少相关的依赖库,安装相应库就解决了：

sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

再次安装,就不再提示了。

(3)环境变量配置

打开~/.bashrc文件：sudo gedit ~/.bashrc将以下内容写入到~/.bashrc尾部：

export CUDA_HOME=/usr/local/cuda-8.0export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64:/usr/local/cuda-8.0/extras/CUPTI/lib64:$LD_LIBRARY_PATHsexport PATH=/usr/local/cuda-8.0/bin:$PATHexport LD_LIBRARY_PATH="/usr/local/cuda-8.0/lib64:/usr/local/cuda-8.0/extras/CUPTI/lib64:/usr/local/cuda-8.0/lib64:/usr/local/cuda-8.0/lib64"

刷新：

source ~/.bashrc

(4)配置cuDNN

注意：首次配置cuDNN时我是下载的cuDNN6.0版本，但是后来在编译fast rcnn及SSD时发现有很多问题是由于cuDNN版本不匹配引起的，因此后来又手动删除掉了cuDNN6.0的include和lib文件，重新下载cuDNN5.1版本并重新编译caffe。

cuDNN是GPU加速计算深层神经网络的库。首先去官网/rdp/cudnn-download 下载cuDNN，需要注册一个账号才能下载。

注册后的下载网址：/rdp/cudnn-archive

在Windows中使用迅雷下载，然后再Ubuntu中使用，这样比较快。

图3.cuDNN下载

下载cuDNN5.1 之后进行解压：

sudo tar -zxvf ./cudnn-8.0-linux-x64-v5.1.tgz

进入cuDNN5.1 解压之后的include目录，在命令行进行如下操作：

cd cuda/includesudo cp cudnn.h /usr/local/cuda/include #复制头文件

再进入lib64目录下的动态文件进行复制和链接：

（这里的libcudnn.so.5.1.10是固有文件，而libcudnn.so.5是libcudnn.so.5.1.10链接得到的动态文件，libcudnn.so是libcudnn.so.5链接得到的动态文件。）

cd ..cd lib64 sudo cp lib* /usr/local/cuda/lib64/ #复制动态链接库

cd /usr/local/cuda/lib64/sudo rm -rf libcudnn.so libcudnn.so.5 #删除原有动态文件

sudo ln -s libcudnn.so.5.1.10 libcudnn.so.5 #生成软衔接（注意这里要和自己下载的cudnn版本对应，可以在/usr/local/cuda/lib64下查看自己libcudnn的版本） sudo ln -s libcudnn.so.5 libcudnn.so #生成软链接

这里需要注意上面这个命令，网上有人的第三行命令为：

sudo ln -s libcudnn.so.5.1.5 libcudnn.so.5 #生成软衔接

起初我执行的也是上条链接 libcudnn.so.5.1.5 的命令，但是后面编译caffe时出错，报错内容为 /usr/bin/ld: 找不到 -lcudnn，所以这里需要先查看一下自己应该链接的是 libcudnn.so.5.1.10 还是 libcudnn.so.5.1.5 ，查看方法为下：

locate libcudnn.so

我执行完后显示如下：

yhao@yhao-X550VB:~$ locate libcudnn.so /home/yhao/.local/share/Trash/files/libcudnn.so /home/yhao/.local/share/Trash/files/libcudnn.so.5 /home/yhao/.local/share/Trash/files/libcudnn.so.5.1.10 /home/yhao/.local/share/Trash/files/cuda/lib64/libcudnn.so /home/yhao/.local/share/Trash/files/cuda/lib64/libcudnn.so.5 /home/yhao/.local/share/Trash/files/cuda/lib64/libcudnn.so.5.1.10 /home/yhao/.local/share/Trash/info/libcudnn.so.5.1.10.trashinfo /home/yhao/.local/share/Trash/info/libcudnn.so.5.trashinfo /home/yhao/.local/share/Trash/info/libcudnn.so.trashinfo /home/yhao/cuda/lib64/libcudnn.so /home/yhao/cuda/lib64/libcudnn.so.5 /home/yhao/cuda/lib64/libcudnn.so.5.1.10 /usr/local/lib/libcudnn.so /usr/local/lib/libcudnn.so.5

可以看到我的文件是 libcudnn.so.5.1.10 ，并没有 libcudnn.so.5.1.5，所以第三行命令我链接的是 libcudnn.so.5.1.10 ，这里第三行链接命令视你的查看结果而定。

注意：下面这个步骤不能缺少！！否则可能链接失败！

执行

sudo ldconfig -v #必须在/usr/local/cuda/lib64/目录下执行，否则可能会报libcudnn.so.5: cannot open shared object file: No such file or directory

或者

sudo ldconfig /usr/local/cuda/lib64

至此，cudnn就配置完成了。

安装完成后可用 nvcc -V 命令验证是否安装成功，若出现以下信息则表示安装成功：

yhao@yhao-X550VB:~$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) - NVIDIA Corporation Built on Tue_Jan_10_13:22:03_CST_ Cuda compilation tools, release 8.0, V8.0.61

(5)测试CUDA的samples

cd /usr/local/cuda-8.0/samples/1_Utilities/deviceQuery sudo make sudo ./deviceQuery

如果显示一些关于GPU的信息，则说明安装成功。

例二：

cd /usr/local/cuda/samples/5_Simulations/fluidsGLsudo make clean && sudo make./fluidsGL

自带例子测试通过，Cuna8.0安装配置完成！

参考：

/jiangyanting/article/details/78873113#commentBox

/CAU_Ayao/article/details/80578600#commentBox

/yhaolpz/article/details/71375762

4.安装OpenCV，这里是在安装anaconda3后，安装的opencv，主要在cmake时，将opencv配置到anaconda3中。

事实证明使用conda便捷安装的opencv是阉割版，不能实现视频和摄像头的读取功能，所以需要自己手动编译。

（1）下载opencv，我在window中用迅雷下载，下载后放到Ubuntu中，这样比较快。

/releases.html

解压后，放在软件安装目录下，我是放在了“/home/zqzy”目录下了

（2）更新，否则可能会面会报错

sudo apt-get updatesudo apt-get upgrade

（3）安装依赖

sudo apt-get install build-essential sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-devsudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-devsudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-devsudo apt-get install libxvidcore-dev libx264-devsudo apt-get install libgtk-3-devsudo apt-get install libatlas-base-dev gfortran pylintsudo apt-get install python2.7-dev python3.5-devsudo apt-getinstallbuild-essential sudo apt-getinstall cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev sudo apt-getinstall python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev # 处理图像所需的包 sudo apt-getinstall libavcodec-dev libavformat-dev libswscale-dev libv4l-dev liblapacke-devsudo apt-getinstall libxvidcore-dev libx264-dev # 处理视频所需的包 sudo apt-getinstall libatlas-base-dev gfortran # 优化opencv功能sudo apt-getinstall ffmpeg

（4）执行如下命令：

cd /home/zqzy/opencv-3.3.0mkdir buildcd build

（5）cmake，这里根据自己的安装目录进行修改

针对anaconda3，进行安装

/*执行时的命令*/cmake -D WITH_IPP=OFF -D PYTHON_DEFAULT_EXECUTABLE=/home/zqzy/anaconda3/bin/python3 -D BUILD_opencv_python3=ON -D BUILD_opencv_python2=OFF -D PYTHON3_EXCUTABLE=/home/zqzy/anaconda3/bin/python3 -D PYTHON3_INCLUDE_DIR=/home/zqzy/anaconda3/include/python3.6m -D PYTHON3_LIBRARY=/home/zqzy/anaconda3/lib/libpython3.6m.so.1.0 -D PYTHON_NUMPY_PATH=/home/zqzy/anaconda3/lib/python3.6/site-packages ..

/* 上面的命令，如下，这样好阅读，但是还是用上面的命令执行cmake -D WITH_IPP=OFF -D PYTHON_DEFAULT_EXECUTABLE=/home/zqzy/anaconda3/bin/python3 -D BUILD_opencv_python3=ON -D BUILD_opencv_python2=OFF -D PYTHON3_EXCUTABLE=/home/zqzy/anaconda3/bin/python3 -D PYTHON3_INCLUDE_DIR=/home/zqzy/anaconda3/include/python3.6m -D PYTHON3_LIBRARY=/home/zqzy/anaconda3/lib/libpython3.6m.so.1.0 -D PYTHON_NUMPY_PATH=/home/zqzy/anaconda3/lib/python3.6/site-packages ..*/

针对anaconda2，进行安装

cmake -D WITH_IPP=OFF -D PYTHON_DEFAULT_EXECUTABLE=/home/zqzy/anaconda2/bin/python2 -D BUILD_opencv_python3=OFF -D BUILD_opencv_python2=ON -D PYTHON2_EXCUTABLE=/home/zqzy/anaconda2/bin/python -D PYTHON2_INCLUDE_DIR=/home/zqzy/anaconda2/include/python2.7 -D PYTHON2_LIBRARY=/home/zqzy/anaconda2/lib/libpython2.7.so.1.0 -D PYTHON_NUMPY_PATH=/home/zqzy/anaconda2/lib/python2.7/site-packages ..

/*上面的命令，如下，这样好阅读，但是还是用上面的命令执行cmake -D WITH_IPP=OFF -D PYTHON_DEFAULT_EXECUTABLE=/home/zqzy/anaconda2/bin/python2 -D BUILD_opencv_python3=OFF -D BUILD_opencv_python2=ON -D PYTHON2_EXCUTABLE=/home/zqzy/anaconda2/bin/python -D PYTHON2_INCLUDE_DIR=/home/zqzy/anaconda2/include/python2.7 -D PYTHON2_LIBRARY=/home/zqzy/anaconda2/lib/libpython2.7.so.1.0 -D PYTHON_NUMPY_PATH=/home/zqzy/anaconda2/lib/python2.7/site-packages ..*/

（6）接下来一步该make了，我用的是，这步时间比较久，耐心等等，如果这步没有报错，那么就离成功不远了.

make -j8

（7）接下来再用几句简单命令即可完成安装：

sudo make installsudo gedit /etc/ld.so.conf.d/opencv.conf

在打开的写字板中输入：

/usr/local/lib

在终端输入：

sudo ldconfig

最后设置路径：

sudo gedit /etc/bash.bashrc

文件末尾添加并保存：

PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfigexport PKG_CONFIG_PATH

测试

终端输入python，import cv2，报错 NO module in cv2，应该是找不到cv2.so文件。

针对anaconda3，进行安装：

解决方法：

去opencv-3.3.0/build/lib/python3这个路径下找到cv2.cpython-36m-x86_64-linux-gnu.so文件，

然后复制到/Software/anaconda3/lib/python3.6/site-packages文件夹下。

cp /home/zqzy/opencv-3.3.0/build/lib/python3/cv2.cpython-36m-x86_64-linux-gnu.so /home/zqzy/anaconda3/lib/python3.6/site-packages

在终端输入python，import cv2，没有报错。

针对anaconda2，进行安装：

解决方法：

去opencv-3.3.0/build/lib这个路径下找到cv2.so文件，

然后复制到/Software/anaconda2/lib/python2.7/site-packages文件夹下。

cp /home/zqzy/opencv-3.3.0-py2/build/lib/cv2.so /home/zqzy/anaconda2/lib/python2.7/site-packages

在终端输入python，import cv2，没有报错。

如果，复制完之后又报错：

ImportError: /home/wz/Software/anaconda3/bin/../lib/libstdc++.so.6: version `GLIBCXX_3.4.21’ not found (required by /home/wz/Software/anaconda3/lib/python3.6/site-packages/cv2.cpython-36m-x86_64-linux-gnu.so)

原因：gcc库版本太老

解决方案： conda install libgcc

应该装好了吧^-^~~

附上测试摄像头的代码

# 打开摄像头并灰度化显示import cv2capture = cv2.VideoCapture(0)while(True):# 获取一帧ret, frame = capture.read()# 将这帧转换为灰度图gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)cv2.imshow('frame', gray)if cv2.waitKey(1) == ord('q'):breakcapture.release()cv2.destroyAllWindows()

参考教程：/aaon22357/article/details/81913465

/releases.html

5.安装caffe，这里是在安装anaconda3后，安装的caffe，主要在安装时，将caffe配置到anaconda3中。

（1）安装依赖

sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler sudo apt-get install --no-install-recommends libboost-all-dev sudo apt-get install libatlas-base-dev sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-devsudo apt-get install build-essential sudo apt-get install python-dev

（2）下载caffe

git clone /BVLC/caffe.git

（3）配置Makefile.config文件之前，先复制几个文件，有的没有“libpython3.6m”这个文件

针对anaconda3，有的anaconda3没有libpython3.6m*，就不需要执行这句话。

sudo cp ~/anconda3/bin/libpython3.6m* /usr/lib/x86_64-linux-gnu

针对 anaconda2，有的anaconda2也没有libpython2.7m*，就不需要执行这句话。

（4）将Makefile.config.example的内容复制到Makefile.config

sudo cp Makefile.config.example Makefile.config

（5）修改Makefile.config文件

sudo gedit Makefile.config #打开Makefile.config文件

（6）修改Makefile.config文件

针对anaconda3，修改Makefile.config文件

a.使用 cudnn USE_CUDNN := 1 b.使用的 opencv 版本是 3 OPENCV_VERSION := 3 c.使用 python 来编写 WITH_PYTHON_LAYER := 1 d.重要的一项 : 将# Whatever else you find you need goes here.下面的 INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib 修改为： INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/ LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial这是因为ubuntu16.04的文件包含位置发生了变化，尤其是需要用到的hdf5的位置，所以需要更改这一路径 e.删除 CUDA_ARCH := 的前两行,避免 CUDA 报错 f. 修改 PYTHON_INCLUDE 路径，将其注释即可# We need to be able to find Python.h and numpy/arrayobject.h. # PYTHON_INCLUDE := /usr/include/python2.7 \# /usr/lib/python2.7/dist-packages/numpy/core/include g. 修改 Anaconda 路径 # Verify anaconda location, sometimes it's in root. ANACONDA_HOME := /home/zqzy/anaconda3 #改成自己anaconda的路径#注意这里的 \，书写方式是“空一个空格，然后回车，下一行使用Tab建进行空格”PYTHON_INCLUDE := $(ANACONDA_HOME)/include \ $(ANACONDA_HOME)/include/python3.6m \$(ANACONDA_HOME)/lib/python3.6/site-packages/numpy/core/include# 注意这里的空格是使用“Tab”键来写的h. 使用 Python3 # Uncomment to use Python 3 (default is Python 2) PYTHON_LIBRARIES := boost_python3 python3.6m #注意这里的 \，书写方式是“空一个空格，然后回车，下一行使用Tab建进行空格”PYTHON_INCLUDE := /usr/include/python3.6m \/usr/lib/python3.6/dist-packages/numpy/core/include \/home/senius/anaconda3/include/python3.6m # 注意这里的空格是使用“Tab”键来写的i. 修改 PYTHON_LIB 路径 # We need to be able to find libpythonX.X.so or .dylib. #PYTHON_LIB := /usr/lib #注意这里的 \，书写方式是“空一个空格，然后回车，下一行使用Tab建进行空格”PYTHON_LIB := $(ANACONDA_HOME)/lib \$(ANACONDA_HOME)/pkgs/python-3.6.4-hc3d631a_1/lib# 注意这里的空格是使用“Tab”键来写的j. 紧接着上面这个PYTHON_LIB，在其下面添加LINKFLAGS := -Wl,-rpath,/home/zqzy/anaconda3/lib #要添加这句话，否则报错以上内容一定要仔细查看自己相应的目录一一对应，若没有相关目录一定是少安装了某些依赖，这一步配置好了后面就不会有什么错误

以上内容一定要仔细查看自己相应的目录一一对应，有的可能不太一样，需要自己去相应的目录查看，若没有相关目录一定是少安装了某些依赖，这一步配置好了后面就不会有什么错误

这里附上我修改后的针对anaconda3的Makefile.config：

## Refer to /installation.html# Contributions simplifying and improving our build system are welcome!# cuDNN acceleration switch (uncomment to build with cuDNN).USE_CUDNN := 1# CPU-only switch (uncomment to build without GPU support).# CPU_ONLY := 1# uncomment to disable IO dependencies and corresponding data layers# USE_OPENCV := 0# USE_LEVELDB := 0# USE_LMDB := 0# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)#You should not set this flag if you will be reading LMDBs with any#possibility of simultaneous read and write# ALLOW_LMDB_NOLOCK := 1# Uncomment if you're using OpenCV 3OPENCV_VERSION := 3# To customize your choice of compiler, uncomment and set the following.# N.B. the default for Linux is g++ and the default for OSX is clang++# CUSTOM_CXX := g++# CUDA directory contains bin/ and lib/ directories that we need.CUDA_DIR := /usr/local/cuda# On Ubuntu 14.04, if cuda tools are installed via# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:# CUDA_DIR := /usr# CUDA architecture setting: going with all of them.# For CUDA < 6.0, comment the *_50 through *_61 lines for compatibility.# For CUDA < 8.0, comment the *_60 and *_61 lines for compatibility.CUDA_ARCH := -gencode arch=compute_30,code=sm_30 \-gencode arch=compute_35,code=sm_35 \-gencode arch=compute_50,code=sm_50 \-gencode arch=compute_52,code=sm_52 \-gencode arch=compute_60,code=sm_60 \-gencode arch=compute_61,code=sm_61 \-gencode arch=compute_61,code=compute_61# BLAS choice:# atlas for ATLAS (default)# mkl for MKL# open for OpenBlasBLAS := atlas# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.# Leave commented to accept the defaults for your choice of BLAS# (which should work)!# BLAS_INCLUDE := /path/to/your/blas# BLAS_LIB := /path/to/your/blas# Homebrew puts openblas in a directory that is not on the standard search path# BLAS_INCLUDE := $(shell brew --prefix openblas)/include# BLAS_LIB := $(shell brew --prefix openblas)/lib# This is required only if you will compile the matlab interface.# MATLAB directory should contain the mex binary in /bin.# MATLAB_DIR := /usr/local# MATLAB_DIR := /Applications/MATLAB_Rb.app# NOTE: this is required only if you will compile the python interface.# We need to be able to find Python.h and numpy/arrayobject.h.# PYTHON_INCLUDE := /usr/include/python2.7 \#/usr/lib/python2.7/dist-packages/numpy/core/include# Anaconda Python distribution is quite popular. Include path:# Verify anaconda location, sometimes it's in root.ANACONDA_HOME := /home/zqzy/anaconda3PYTHON_INCLUDE := $(ANACONDA_HOME)/include \$(ANACONDA_HOME)/include/python3.6m \$(ANACONDA_HOME)/lib/python3.6/site-packages/numpy/core/include# Uncomment to use Python 3 (default is Python 2)PYTHON_LIBRARIES := boost_python3 python3.6mPYTHON_INCLUDE := /usr/include/python3.6m \/usr/lib/python3.6/dist-packages/numpy/core/include \/home/zqzy/anaconda3/include/python3.6m# We need to be able to find libpythonX.X.so or .dylib.#PYTHON_LIB := /usr/libPYTHON_LIB := $(ANACONDA_HOME)/lib \$(ANACONDA_HOME)/pkgs/python-3.6.4-hc3d631a_1/libLINKFLAGS := -Wl,-rpath,/home/zqzy/anaconda3/lib# Homebrew installs numpy in a non standard path (keg only)# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include# PYTHON_LIB += $(shell brew --prefix numpy)/lib# Uncomment to support layers written in Python (will link against Python libs)WITH_PYTHON_LAYER := 1# Whatever else you find you need goes here.INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serialLIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies# INCLUDE_DIRS += $(shell brew --prefix)/include# LIBRARY_DIRS += $(shell brew --prefix)/lib# NCCL acceleration switch (uncomment to build with NCCL)# /NVIDIA/nccl (last tested version: v1.2.3-1+cuda8.0)# USE_NCCL := 1# Uncomment to use `pkg-config` to specify OpenCV library paths.# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)# USE_PKG_CONFIG := 1# N.B. both build and distribute dirs are cleared on `make clean`BUILD_DIR := buildDISTRIBUTE_DIR := distribute# Uncomment for debugging. Does not work on OSX due to /BVLC/caffe/issues/171# DEBUG := 1# The ID of the GPU that 'make runtest' will use to run unit tests.TEST_GPUID := 0# enable pretty build (comment to see full commands)Q ?= @

针对anaconda2，修改Makefile.config文件

a.使用 cudnn USE_CUDNN := 1 b.使用的 opencv 版本是 3 OPENCV_VERSION := 3 c.使用 python 来编写 WITH_PYTHON_LAYER := 1 d.重要的一项 : 将# Whatever else you find you need goes here.下面的 INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib 修改为： INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/ LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial这是因为ubuntu16.04的文件包含位置发生了变化，尤其是需要用到的hdf5的位置，所以需要更改这一路径 e.删除 CUDA_ARCH := 的前两行,避免 CUDA 报错 f. 修改 PYTHON_INCLUDE 路径，将其注释即可# We need to be able to find Python.h and numpy/arrayobject.h. # PYTHON_INCLUDE := /usr/include/python2.7 \# /usr/lib/python2.7/dist-packages/numpy/core/include g. 修改 Anaconda 路径 # Anaconda Python distribution is quite popular. Include path:# Verify anaconda location, sometimes it's in root.ANACONDA_HOME := /home/zqzy/anaconda2 #取消注释，并且修改为自己anaconda的安装路径PYTHON_INCLUDE := $(ANACONDA_HOME)/include \#取消注释$(ANACONDA_HOME)/include/python2.7 \ #取消注释，注意这里空格是Tab键$(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include ##取消注释# 注意这里的空格是使用“Tab”键来写的h. 修改 PYTHON_LIB 路径 # We need to be able to find libpythonX.X.so or .dylib. #PYTHON_LIB := /usr/lib #注意这里的 \，书写方式是“空一个空格，然后回车，下一行使用Tab建进行空格”PYTHON_LIB := $(ANACONDA_HOME)/lib \$(ANACONDA_HOME)/pkgs/python-2.7.14-h1571d57_29/lib# 注意这里的空格是使用“Tab”键来写的i. 紧接着上面这个PYTHON_LIB，在其下面添加LINKFLAGS := -Wl,-rpath,/home/zqzy/anaconda2/lib #要添加这句话，否则报错以上内容一定要仔细查看自己相应的目录一一对应，若没有相关目录一定是少安装了某些依赖，这一步配置好了后面就不会有什么错误

这里附上我修改后的针对anaconda2的Makefile.config：

## Refer to /installation.html# Contributions simplifying and improving our build system are welcome!# cuDNN acceleration switch (uncomment to build with cuDNN).USE_CUDNN := 1# CPU-only switch (uncomment to build without GPU support).# CPU_ONLY := 1# uncomment to disable IO dependencies and corresponding data layers# USE_OPENCV := 0# USE_LEVELDB := 0# USE_LMDB := 0# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)#You should not set this flag if you will be reading LMDBs with any#possibility of simultaneous read and write# ALLOW_LMDB_NOLOCK := 1# Uncomment if you're using OpenCV 3OPENCV_VERSION := 3# To customize your choice of compiler, uncomment and set the following.# N.B. the default for Linux is g++ and the default for OSX is clang++# CUSTOM_CXX := g++# CUDA directory contains bin/ and lib/ directories that we need.CUDA_DIR := /usr/local/cuda# On Ubuntu 14.04, if cuda tools are installed via# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:# CUDA_DIR := /usr# CUDA architecture setting: going with all of them.# For CUDA < 6.0, comment the *_50 through *_61 lines for compatibility.# For CUDA < 8.0, comment the *_60 and *_61 lines for compatibility.CUDA_ARCH := -gencode arch=compute_30,code=sm_30 \-gencode arch=compute_35,code=sm_35 \-gencode arch=compute_50,code=sm_50 \-gencode arch=compute_52,code=sm_52 \-gencode arch=compute_60,code=sm_60 \-gencode arch=compute_61,code=sm_61 \-gencode arch=compute_61,code=compute_61# BLAS choice:# atlas for ATLAS (default)# mkl for MKL# open for OpenBlasBLAS := atlas# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.# Leave commented to accept the defaults for your choice of BLAS# (which should work)!# BLAS_INCLUDE := /path/to/your/blas# BLAS_LIB := /path/to/your/blas# Homebrew puts openblas in a directory that is not on the standard search path# BLAS_INCLUDE := $(shell brew --prefix openblas)/include# BLAS_LIB := $(shell brew --prefix openblas)/lib# This is required only if you will compile the matlab interface.# MATLAB directory should contain the mex binary in /bin.# MATLAB_DIR := /usr/local# MATLAB_DIR := /Applications/MATLAB_Rb.app# NOTE: this is required only if you will compile the python interface.# We need to be able to find Python.h and numpy/arrayobject.h.# PYTHON_INCLUDE := /usr/include/python2.7 \#/usr/lib/python2.7/dist-packages/numpy/core/include# Anaconda Python distribution is quite popular. Include path:# Verify anaconda location, sometimes it's in root.ANACONDA_HOME := /home/zqzy/anaconda2PYTHON_INCLUDE := $(ANACONDA_HOME)/include \$(ANACONDA_HOME)/include/python2.7 \$(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include# Uncomment to use Python 3 (default is Python 2)# PYTHON_LIBRARIES := boost_python3 python3.5m# PYTHON_INCLUDE := /usr/include/python3.5m \# /usr/lib/python3.5/dist-packages/numpy/core/include# We need to be able to find libpythonX.X.so or .dylib.# PYTHON_LIB := /usr/libPYTHON_LIB := $(ANACONDA_HOME)/lib \$(ANACONDA_HOME)/pkgs/python-2.7.14-h1571d57_29/libLINKFLAGS := -Wl,-rpath,/home/zqzy/anaconda2/lib# Homebrew installs numpy in a non standard path (keg only)# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include# PYTHON_LIB += $(shell brew --prefix numpy)/lib# Uncomment to support layers written in Python (will link against Python libs)WITH_PYTHON_LAYER := 1# Whatever else you find you need goes here.# INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include# LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/libINCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/ LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies# INCLUDE_DIRS += $(shell brew --prefix)/include# LIBRARY_DIRS += $(shell brew --prefix)/lib# NCCL acceleration switch (uncomment to build with NCCL)# /NVIDIA/nccl (last tested version: v1.2.3-1+cuda8.0)# USE_NCCL := 1# Uncomment to use `pkg-config` to specify OpenCV library paths.# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)# USE_PKG_CONFIG := 1# N.B. both build and distribute dirs are cleared on `make clean`BUILD_DIR := buildDISTRIBUTE_DIR := distribute# Uncomment for debugging. Does not work on OSX due to /BVLC/caffe/issues/171# DEBUG := 1# The ID of the GPU that 'make runtest' will use to run unit tests.TEST_GPUID := 0# enable pretty build (comment to see full commands)Q ?= @

（6）修改 Makefile 文件把下面第一行代码改为第二行代码：

将：LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_hl hdf5 替换为：LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial opencv_core opencv_imgproc opencv_imgcodecs opencv_highgui

（7）修改 Makefile 文件把下面第一行代码改为第二行代码：

将： NVCCFLAGS +=-ccbin=$(CXX) -Xcompiler-fPIC $(COMMON_FLAGS) 替换为： NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)

（8）修改 Makefile 文件把下面第一段代码改为第二段代码：

将：

# Debuggingifeq ($(DEBUG), 1)COMMON_FLAGS += -DDEBUG -g -O0NVCCFLAGS += -GelseCOMMON_FLAGS += -DNDEBUG -O2endif

修改为：

# Debuggingifeq ($(DEBUG), 1)COMMON_FLAGS += -DDEBUG -g -O0NVCCFLAGS += -GelseCOMMON_FLAGS += -DNDEBUG -O2NVCCFLAGS += -Gendif# 注意这里的空格是使用“Tab”键来写的。

也就是加一句 NVCCFLAGS += -G

（9）针对anaconda3

使用“locatelibpython3.6m.so.1.0” 查找 “libpython3.6m.so.1.0 ” 的位置，查找出来是在anaconda3/lib中

执行如下命令：

sudo gedit /etc/ld.so.conf/home/zqzy/anaconda3/lib/sudo ldconfig

针对anaconda2

使用“locatelibpython2.7.so.1.0” 查找 “libpython2.7.so.1.0 ” 的位置，查找出来是在/usr/lib/x86_64-linux-gnu/中

执行如下命令：

执行命令：sudo gedit /etc/ld.so.conf添加内容：/usr/lib/x86_64-linux-gnu/执行命令：sudo ldconfig

（10）编辑 /usr/local/cuda/include/host_config.h ，将其中的第115行注释掉

//#error -- unsupported GNU version! gcc versions later than 5 are not supported!

（11）这里只针对anaconda3，对于anaconda2不需要

因为在配置文件的第 74 行中有PYTHON_LIBRARIES := boost_python3，但是在系统中无法找到boost_python3.lib这个库文件。

解决方案如下：

检查是否有如下文件：

ls /usr/lib/x86_64-linux-gnu/libboost_python-py35.so

这里如果有，说明我们的系统中已经有了这种库文件，只是文件名不同。

添加软链接

cd /usr/lib/x86_64-linux-gnu/sudo ln -s libboost_python-py35.so libboost_python3.sosudo ln -s libboost_python-py35.a libboost_python3.a sudo ln -s libboost_python-py35.so.1.58.0 libboost_python3.so.1.58.0 然后 sudo gedit /etc/ld.so.conf 添加 /usr/lib/x86_64-linux-gnu sudo ldconfig

针对anaconda2，不需要第11这一步，因为没有在配置文件的第 74 行中添加PYTHON_LIBRARIES := boost_python3

（12）将cuda信息，复制到“/usr/local/lib”的相应路径下

sudo cp /usr/local/cuda-8.0/lib64/libcudart.so.8.0 /usr/local/lib/libcudart.so.8.0 && sudo ldconfigsudo cp /usr/local/cuda-8.0/lib64/libcublas.so.8.0 /usr/local/lib/libcublas.so.8.0 && sudo ldconfig sudo cp /usr/local/cuda-8.0/lib64/libcurand.so.8.0 /usr/local/lib/libcurand.so.8.0 && sudo ldconfig sudo cp /usr/local/cuda-8.0/lib64/libcudnn.so.5 /usr/local/lib/libcudnn.so.5 && sudo ldconfig #这里根据自己编译运行时的提示错误，进行相应的拷贝，有的是libcudnn.so.5 ，有的是libcudnn.so.6

（13）编译

cd .. #到caffe-master目录sudo make all -j8sudo make pycaffe -j8

（14）添加环境变量

sudo gedit ~/.bashrc最后一行添加export PYTHONPATH=/home/zqzy/caffe-master/python:$PYTHONPATHsource ~/.bashrc

（15）安装protobuf

先修改权限，否则安装不了：

sudo chown -R zqzy:zqzy /home/zqzy/anaconda3#这里的zqzy是我的用户名

在Anaconda终端搜索protobuf库，如图：

然后安装protobuf（选择你对应的版本，我这里是python3.6，所以安装protobuf是3.2.0），如图：

（16）测试

sudo make runtest -j8

（16）在Python中导入，没有错误，即可