Caffe环境安装

一、安装依赖包：

　　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

二、安装protobuf

       1.删除系统原生protobuf

　　　　sudo rm -rf protoc

       2.确认当前protobuf是否已删除

　　3.安装protobuf2.6.1

            cd protobuf2.6.1

　　　./configure

　　　make

　　　sudo make install

           sudo ldconfig

　　4.确认当前protobuf版本

三、安装opencv3.4.6

　　mkdir build && cd build # 创建编译的文件目录,并进入该目录

       cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..

　　注：在docker环境下安装时，碰到一个问题：/usr/bin/ld: warning: libzstd.so.1.3.7, needed by //opt/conda/lib/libtiff.so.5, not found (try using -rpath or -rpath-link)

　　解决方法：加上选项：-D BUILD_TIFF=ON

　　 cmake -D BUILD_TIFF=ON -D CMAKE_BUILD_TYPE=Release -D ENABLE_PRECOMPILED_HEADERS=OFF -D CMAKE_INSTALL_PREFIX=/usr/local ..

　　make -j8  #编译

　　sudo make install #安装

　　pkg-config --modversion opencv  #验证成功与否

四、安装依赖包

　　1)sudo apt-get install libhdf5-dev

五、安装caffe

　　sudo cp Makefile.config.example Makefile.config

1、修改Makefile.config
       sudo gedit Makefile.config
       应用 opencv 版本将
       #OPENCV_VERSION := 3   修改为： OPENCV_VERSION := 3
       修改 python 路径
       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     

2、修改Makefie

       修改 caffe 目录下的 Makefile 文件
       将：NVCCFLAGS +=-ccbin=$(CXX) -Xcompiler-fPIC $(COMMON_FLAGS)
       替换为：NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)
       将：LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_hl hdf5
       改为：LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial

3、编译caffe

       make all -j8

       问题１： 在运行上面命令编译时，报错：//opt/conda/lib/libpng16.so.16: undefined reference to `inflateValidate@ZLIB_1.2.9

       原因:zlib的版本不兼容导致，这里指出是ZLIB1.2.9，本地安装的是zlib1.2.8

       解决方法：

wget http://www.zlib.net/fossils/zlib-1.2.9.tar.gz
tar -xvf zlib-1.2.9.tar.gz
cd zlib-1.2.9
mkdir build
cd build
cmake ..
make
make install

      sudo make runtest -j8  #测试caffe安装是否成功

      sudo apt-get install python3.8-dev

      sudo make pycaffe-j8   #生成最终版本

4.运行caffe

问题１：  import caffe时报错：

　　Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/li.guangyao/Programming/caffe/python/caffe/__init__.py", line 1, in <module> from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, AdaDeltaSolver, AdamSolver, NCCL, Timer File "/home/li.guangyao/Programming/caffe/python/caffe/pycaffe.py", line 13, in <module> from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, ImportError: /home/li.guangyao/Programming/caffe/python/caffe/_caffe.so:

　undefined symbol: _ZN5boost6python6detail11init_moduleER11PyModuleDefPFvvE

解决方法：

　先使用以下命令，安装最新版本libboost: sudo apt-get install --no-install-recommends libboost-all-dev

    打开caffe目录下Makefile文件：

将PYTHON_LIBRARIES := boost_python python2.7
改PYTHON_LIBRARIES := boost_python-py36 #py36需要改为你自己的python版本，如py34,py33

 问题２：修改完问题１，指定PYTHON_LIBRARIES版本为3.6，编译make pycaffe时又报错：/usr/bin/ld: cannot find -lboost_python-py36

 解决方法：

　　１．下载包boost_1_67_0，进入boost源文件目录boost_1_67_0

　　２．运行以下两条命令：

./bootstrap.sh --with-libraries=python --with-toolset=gcc
./b2 cflags='-fPIC' cxxflags='-fPIC' --with-python include="/qp/3rd/anaconda3/include/python3.6m/"
　３．sudo ./b2 install

编译安装成功后，/usr/local/lib下会有libboost_python36.so和libboost_python36.a，有些应用link时需要的是libboost_python3.so或者libboost_python3.a，我们建个软链：

cd /usr/local/lib

sudo ln -s libboost_python-py36.so libboost_python3.so
sudo ln -s libboost_python-py36.a libboost_python3.a

问题３：ModuleNotFoundError: No module named 'skimage'
解决方法：pip install scikit-image

问题４：ModuleNotFoundError: No module named 'google'
解决方法：pip install google
        pip install protobuf