Caffe Ubuntu16.04环境配置

Caffe

环境

官网最新版Ubuntu16.04.3

CPU-Only

依照教程，执行如下过程

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

安装Anaconda的Python3.6 version。
强烈推荐vscode进行文本编辑。

补充依赖包

网盘中有glog包，下载后

tar zxvf glog-0.3.3.tar.gz
cd glog-0.3.3
./configure
sudo make 
sudo make install

同时，依据Ubuntu16.04安装caffe中遇到的问题总结，命令行输入如下命令继续安装相关依赖包

#1. ./include/caffe/common.hpp:5:27: fatal error: gflags/gflags.h: No such file or directory
解决办法：sudo apt-get install libgflags-dev
#2.  ./include/caffe/util/mkl_alternate.hpp:14:19: fatal error: cblas.h: No such file or directory
解决办法：sudo apt-get install libblas-dev
#3. ./include/caffe/util/db_lmdb.hpp:8:18: fatal error: lmdb.h: No such file or directory
解决办法：sudo apt install liblmdb-dev
#4. /usr/bin/ld: cannot find -lcblas
    /usr/bin/ld: cannot find -latlas
解决办法：sudo apt install libatlas-base-dev

其中关于HDF5的问题，不是按照博客原来的方式解决的，博客原来采用这种方式

#3. ./include/caffe/util/hdf5.hpp:6:18: fatal error: hdf5.h: No such file or directory
解决办法：在Makefile.config找到以下行并添加蓝色部分
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/hdf5/serial

但是这个在进入后面make runtest的时候会出现HDF5的相关问题，无法通过增加软链接来解决。这里参照了另一个办法并进行了细微的修改，去官网下载。
下载解压后，输入命令

./configure --prefix=/usr/local/lib/hdf5-1.8.3/
make ##4分钟左右
sudo make install

这样就直接把hdf5放入了lib的常用位置。但是这个还是跑不通测试，最终在这里找到了办法,anaconda3里面自带了这个库，所以只用添加进去就好。

编译

进入caffe文件夹，输入

cp Makefile.config.example Makefile.config
code Makefile.config

对文件进行了一些修改，针对CPU模式，主要是Anaconda Python的一些参数调整

# CPU-only switch (uncomment to build without GPU support).
CPU_ONLY := 1
# Uncomment if you're using OpenCV 3
# OPENCV_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++
# 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)/anaconda3
PYTHON_INCLUDE := $(ANACONDA_HOME)/include \
        $(ANACONDA_HOME)/include/python3.6m \
		$(ANACONDA_HOME)/lib/python3.6/site-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
# Uncomment to support layers written in Python (will link against Python libs)
WITH_PYTHON_LAYER := 1
# Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171
DEBUG := 1

之后就是

make all
make test
make runtest

可以全部通过，没有问题。

安装Opencv

最核心的部分安装成功后，安装opencv支持。

sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev

百度网盘下载准备好opencv和opencv_contrib两个文件夹，而后

cd opencv
mkdir build
cd build
cmake –D CMAKE_BUILD_TYPE=Release –D OPENCV_EXTRA_MODULES_PATH=/home/banshee/opencv_contrib/modules/ -D CMAKE_INSTALL_PREFIX=/usr/local –D WITH_IPP=OFF ..

在这个过程中，会遇到ippicv_linux_20151201.tgz国内被墙了，有时候下载不成功或者提出hash值不对

此时果断中止，然后下载对应文件到对应目录即可（/3rdparty/ippicv/downloads/linux-808b791a6eac9ed78d32a7666804320e/ippicv_linux_20151201.tgz）
而后

（<opencv目录>/3rdparty/ippicv/downloads/linux-808b791a6eac9ed78d32a7666804320e/ippicv_linux_20151201.tgz）

而后

sudo make -j4
sudo make install

然后运行测试代码

#include <iostream>
#include <opencv2/highgui/highgui.hpp>
int main(int argc,char* argv[]){
  const std::string window_name = "lena";
  const std::string input_pic   = "lena.jpg";
  cv::Mat test_pic = cv::imread(input_pic);
  if(test_pic.empty()){
      std::cout<<"no input image"<<std::endl;
      return 1;
  }
  cv::namedWindow(window_name);
  cv::imshow(window_name,test_pic);
  cv::waitKey(0);
  return 0;
}

对应的测试图片lena.jpg在网盘中也有。

g++ -o test_opencv test_opencv.cpp -lopencv_core -lopencv_highgui -lopencv_imgcodecs

完成后，重新编译caffe

make all
make test
make runtest
make pycaffe

Pycaffe

在上文编译好pycaffe后，我们会发现pycaffe并不能使用，会出现如下报错

ImportError: /home/zero/Documents/caffe-master/python/caffe/_caffe.so:undefined symbol: 
_ZN5boost6python6detail11init_moduleER11PyModuleDefPFvvE

这个bug是由boost版本不匹配导致的，涉及到Makefile中的这样一行

PYTHON_LIBRARIES ?= boost_python python2.7

为了方便，笔者将它修改成了

PYTHON_LIBRARIES := boost_python3 python3.5m

并且增加了如下操作：

cd /usr/lib/x86_64-linux-gnu/
sudo ln -s libboost_python-py35.so libboost_python3.so
sudo ln -s libpython3.5m.so.1.0 libpython3.5m.so

后面才发现，原来在Makefile.config里面就有一模一样的这句话，可见其实只要取消其中的注释，使用上面三行命令并重新编译即可。

GPU