深度学习之MAC(cpu-only)下利用anaconda安装caffe_mac caffe框架怎么下载-优快云博客

转载自http://blog.youkuaiyun.com/ddreaming/article/details/52739893

如何在Mac下安装caffe

在安装caffe之前，应该了解些计算机的基本知识，以及Linux的基本知识，这是因为Linux和Mac的操作系统非常像。比如什么是Linux,它是怎么来的？Linux和GNU啥关系？gcc又是啥。。Linux的目录配置是咋回事，也就是下图中每个文件都存啥知道不？vim是啥，咋用？shell脚本文件是啥？bash和他啥关系？什么是系统的环境变量？怎么设置环境变量？在mac下怎么通过Homebrew下载软件？如果这些你都很清楚的话，那么直接看第二部分的caffe安装步骤就好。如果你跟我一样，对这些概念稀里糊涂，不太明白，那么建议你在安装caffe之前，用一到两天时间补一补这些基本知识，否则你在安装时候只能是一头雾水。。。关于Linux的基本知识，可以参考鸟哥的linux私房菜，写的非常通俗易懂。我也总结了些对于安装caffe直接相关的概念，大家可以参考第一部分的关于linux的介绍。

一、安装caffe步骤

我的电脑配置如下：

MacBook Pro -OS X EI Captian系统-8G内存-Intel Iris Graphics 6100 1536 MB显卡。

由于电脑配置的不是英伟达显卡，所以不能使用CUDA加速了，只能安装个CPU模式玩一玩。下面是安装的想象步骤。

1.安装caffe需要的依赖包。

opencv2.4:

Anaconda Python 2.7:

snappy:

leveled:

flags:

blog:

ship:

lmdv:

protobuf:

boost:

哈哈，看着很多吧，但是不要怕，下载这些东东，在Mac终端里，只需要几个brew指令而已。所以，如果你电脑里没有Homebrew,那么你要先把Homebrew下载下来。然后我们下面要做的，就是将这些依赖包都下载下来，还需要配置些环境变量。

1.我们先下载Anaconda Python 2.7。下载下来的是个Anaconda2-4.2.0-MacOSX-x86_64.sh文件，我们将这个文件复制到/usr/local/Cellar下面（之所以这么做，是因为使用brew指令下载的东东都会默认存到则个路径，我为了方便，就将所有caffe需要的包都放在这里，完全个人喜好～）。也就是下图中带红点的文件。下面我进行Anaconda的安装。在命令行中输入下面指令

[html] view plain copy

print ?

bash Anaconda2-4.2.0-MacOSX-x86_64.sh

bash Anaconda2-4.2.0-MacOSX-x86_64.sh

执行完成后会生成下图中绿色标注的anaconda2文件。

之后我们设置下环境变量，就是将安装的ananconda2的bin文件的路径添加到PATH中，再将anaconda2的路径添加到DYLD_FALLBACK_LIBARY_PATH中去，指向下面两个命令后第一步就完成了（ps：可以使用set指令查看是否添加成功）。

[html] view plain copy

print ?

<pre name=“code” class=“html”>export PATH=/usr/local/ananconda2/bin:PATH   </li><li class="">export DYLD_FALLBACK_LIBRARY_PATH=/usr/local/Cellar/anaconda2:/usr/local/lib:/usr/lib  </li></ol></div><pre code_snippet_id="1914013" snippet_file_name="blog_20161005_2_5155753" name="code" class="html" style="display: none;"><pre name="code" class="html">export PATH=/usr/local/ananconda2/bin:PATH export DYLD_FALLBACK_LIBRARY_PATH=/usr/local/Cellar/anaconda2:/usr/local/lib:/usr/lib
2.下面进入批量安装的阶段。

安装snappy,leveled,flags,blog,ship,lmdv,opencv..

[html] view plain copy

print ?

brew install –fresh -vd snappy leveldb gflags glog szip lmdb homebrew/science/opencv

brew install --fresh -vd snappy leveldb gflags glog szip lmdb homebrew/science/opencv

安装protobuf以及boost

[html] view plain copy

print ?

brew install –build-from-source –with-python –fresh -vd protobuf
 brew install –build-from-source –fresh -vd boost boost-python

brew install --build-from-source --with-python --fresh -vd protobuf
 brew install --build-from-source --fresh -vd boost boost-python

执行完这些指令，要保证这些文件都出现在/usr/local/Cellar中。如果缺失，那么就用brew指令单独下载。
3.下载caffe源码

[html] view plain copy

print ?

git clone https://github.com/BVLC/caffe.git

git clone https://github.com/BVLC/caffe.git

跟前面一样，我把下载的caffe文件夹复制到了/usr/local/Cellar路径下面。

4.下面我们要写一个Makefile.config文件，这个文件用来生成makefile文件的。在caffe包中给了一个例子，叫做Makefile.config.example，我们就是要根据自己的电脑的情况，重新写一个Makefile.config，所谓重新写，也就是在Makefile.config.example的基础上，删减些注释、改改路径而已。所以就直接复制Makefile.config.example，然后粘贴生成Makefile.config文件，用下面指令可以实现。

[html] view plain copy

print ?

cd /usr/local/Cellar/caffe
cp Makefile.config.example Makefile.config

cd /usr/local/Cellar/caffe
cp Makefile.config.example Makefile.config

5.编译安装caffe要用到我们前几步下载的依赖包，那么每个电脑的情况不一样，这些依赖包的路径也就不一样。所以我们要让caffe找到咱们自己下载的那些依赖包。那么我们怎么告诉caffe呢？这就是靠Makefile.config文件啦，系统会根据这个文件生成Makefile文件，然后当我们执行make指令时，系统就会安装Makefile里面的步骤执行编译安装动作。所以，我们要进入Makefile.config文件夹，按照下面修改：

[html] view plain copy

print ?

MacBook-Pro:caffe wei $ vim Makefile.config   </li><li class="alt">  </li><li class=""># NOTE: this is required only if you will compile the python interface.  </li><li class="alt"># We need to be able to find Python.h and numpy/arrayobject.h.  </li><li class="">PYTHON_INCLUDE := /usr/include/python2.7 \&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;nbsp; </li><li class="alt">                /usr/lib/python2.7/dist-packages/numpy/core/include  </li><li class=""># Anaconda Python distribution is quite popular. Include path:  </li><li class="alt"># Verify anaconda location, sometimes it's in root.  </li><li class="">  <span style="color:#ff0000;">ANACONDA_HOME := /usr/local/Cellar/anaconda2  </li><li class="alt">  PYTHON_INCLUDE := $ (ANACONDA_HOME)/include \
$(ANACONDA_HOME)/include/python2.7 \&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;nbsp; </li><li class="alt">                  $ (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/lib
PYTHON_LIB = $(ANACONDA_HOME)/lib  </li><li class="alt"></span>  </li><li class=""># Homebrew installs numpy in a non standard path (keg only)  </li><li class="alt"># PYTHON_INCLUDE += $ (dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include  </li><li class=""># 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.
</li><li class="alt">LIBRARY_DIRS := $ (PYTHON_LIB) /usr/local/lib /usr/lib
# 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  </li><li class="alt"># LIBRARY_DIRS += $ (shell brew –prefix)/lib
# 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 := build
DISTRIBUTE_DIR := distribute
# Uncomment for debugging. Does not work on OSX due to https://github.com/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 ?= @

<span style=”font-size:18px;”> 

MacBook-Pro:caffe wei$ vim Makefile.config

# 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 := /usr/local/Cellar/anaconda2 PYTHON_INCLUDE :=

(ANACONDAHOME)/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/lib PYTHON_LIB := $(ANACONDA_HOME)/lib # Homebrew installs numpy in a non standard path (keg only) # PYTHON_INCLUDE +=

(dir $(dir$ (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 :=

(PYTHONINCLUDE)/usr/local/includeLIBRARYDIRS:= $(PYTHON_INCLUDE) /usr/local/include LIBRARY_DIRS :=$ (PYTHON_LIB) /usr/local/lib /usr/lib # 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 # 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 := build DISTRIBUTE_DIR := distribute # Uncomment for debugging. Does not work on OSX due to https://github.com/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 ?= @ 要特别注意上面红色标注的路径，这是些路径就只想了我们自己下载的那些工具包,直接影响成败啊，其实主要是指向了anaconda
更改完成后，保存退出（先按esc，在输入冒号：，在输入wq ‘esc’+’:’+’wq’)

6.然后就是make啦，让电脑自动编译安装吧

[html] view plain copy

print ?

make all

make all

这步应该没啥问题。

7.下一步我们要安装caffe的python接口，也就是要编译下pycaffe。执行下面指令：

[html] view plain copy

print ?

for req in $(cat python/requirements.txt); do pip install $ req; done
make pycaffe
make distribute

for req in $(cat python/requirements.txt); do pip install $req; done
make pycaffe
make distribute

8.设置环境变量PYTHONPATH。执行下面指令

[html] view plain copy

print ?

export PYTHONPATH=/usr/local/Cellar/caffe/python

export PYTHONPATH=/usr/local/Cellar/caffe/python

9.进入python看一看～

[html] view plain copy

print ?

cd /usr/local/Cellar/caffe/python

cd /usr/local/Cellar/caffe/python

[html] view plain copy

print ?

python

python

应该显示如下信息：

[html] view plain copy

print ?

Python 2.7.12 |Anaconda 4.2.0 (x86_64)| (default, Jul 2 2016, 17:43:17)
[GCC 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00)] on darwin
Type “help”, “copyright”, “credits” or “license” for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://anaconda.org

Python 2.7.12 |Anaconda 4.2.0 (x86_64)| (default, Jul  2 2016, 17:43:17) 
[GCC 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://anaconda.org

如果显示这些，那就说明成功啦。我们就可以跑一个例子看看啦。

10.运行MNIST例子。运行这个例子，我们要要下载数据集啊，执行下面指令

[html] view plain copy

print ?

cd /usr/local/Cellar/caffe

cd /usr/local/Cellar/caffe

[html] view plain copy

print ?

./data/mnist/get_mnist.sh
./examples/mnist/create_mnist.sh

./data/mnist/get_mnist.sh
./examples/mnist/create_mnist.sh

ps,执行上面两条指令，需要mac电脑里有 wget ，如果没有就会报错，说找不到这个指令，我们可以brew下载

[html] view plain copy

print ?

brew install wget

<span style="color:#ff0000;">brew install wget</span>

下载完成后，如果成功了，会在/usr/local/Cellar/caffe/examples/mnist下生成两个文件夹，下图用红点标注了，这是数据生成的lmdb数据集。

ps:当喔执行

[html] view plain copy

print ?

./examples/mnist/create_mnist.sh

./examples/mnist/create_mnist.sh

指令时，出了点问题，提示我找不到hdf5开头的一个动态链接库。安装相关说明，hdf5是包含在anaconda里面的，而且我在anaconda的lib中也发现了hdf5的动态链接库，如下图：

但是不知道为什么找不到，所以我又用brew下载了一个hdf5，指令如下：

[html] view plain copy

print ?

brew install hdf5

brew install hdf5

然后将anaconda文件夹lib中hdf5开头的链接库都粘贴到下面hdf5文件夹中的lib里面。

这样做之后，就没问题了。希望小伙伴不要遇到这个问题。

11.起飞前最后准备。修改成CPU模式。通过vim进入./examples/mnist/lenet_solver.prototxt。

然后将solver_mode改成CPU模式

[html] view plain copy

print ?

solver_mode: CPU

<span style="font-size:18px;">solver_mode: CPU</span>

然后，可以飞起来了

[html] view plain copy

print ?

cd /usr/local/Cellar/caffe/
./examples/mnist/train_lenet.sh

cd /usr/local/Cellar/caffe/
./examples/mnist/train_lenet.sh

12.运行结果如下

参考资源：

http://www.megastormsystems.com/news/how-to-install-caffe-on-mac-os-x-10-11

http://caffe.berkeleyvision.org/install_osx.html

http://blog.youkuaiyun.com/lkj345/article/details/51260811