[========]
caffe是一个训练卷积神经网络的工具,其能够非常好的支持CNN,最新版本整合了RNN。caffe提供python和matlab的借口,能够简单有效地进行编程和使用。
网上教程很多。
在已经安装好cuda和python以及各种常用的python依赖包之后,caffe在编译之前只需要安装以下依赖库:
- protobuf
- snappy
- leveldb
- openCV
- boost
- lbdb
- gflags
- glog
- hdf5
- openblas, atlas, mkl三选一
Install protobuf:
版本:2.5.0
安装python版
下载完成之后
$ cd protobuf/
$ cd python
$ python setup.py build
$ python setup.py install --user
$ cd ~/temp/
$ git clone https://github.com/google/protobuf.git
$ cd protobuf/
$ ./autogen.sh
$ ./configure --prefix=$HOME/local
$ make
$ make install
Install snappy:
$ cd ~/temp/
$ git clone https://github.com/google/snappy.git
$ cd snappy
$ ./autogen.sh
$ ./configure --prefix=$HOME/local
$ make
$ make install
Install leveldb:
$ cd ~/temp/
$ git clone https://github.com/google/leveldb.git
$ cd leveldb
$ make
$ cp -av libleveldb.* $HOME/local/lib/
$ cp -av include/leveldb $HOME/local/include/
最后两步可以直接手动拷贝
Install OpenCV:
版本:2.4.8
http://stackoverflow.com/questions/28010399/build-opencv-with-cuda-support
$ cd ~/temp/
$ wget 'https://github.com/Itseez/opencv/archive/2.4.8.tar.gz'
$ tar xzf 2.4.8.tar.gz
$ cd opencv-2.4.8/
$ cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=$HOME/local -D BUILD_opencv_gpu=OFF -D CUDA_GENETATION=AUTO ..
$ make
$ make install
Install Boost:
版本:1.55.0
$ cd ~/temp/
$ wget “http://sourceforge.net/projects/boost/files/boost/1.55.0/boost_1_55_0.tar.gz”
$ tar xzf boost_1_55_0.tar.gz
$ cd boost_1_55_0
($ ./bootstrap.sh --prefix=$HOME/local)
$ ./bootstrap.sh --prefix=$HOME/local --with-python=python2.7
$ ./b2 -j 32
$ ./b2 install
...failed updating 2 targets...
...skipped 6 targets...
...updated 62 targets...
Install lmdb:
$ cd ~/temp
$ git clone https://github.com/LMDB/lmdb
$ cd mdb/libraries/liblmdb
$ make
$ mkdir $HOME/local/man/man1
$ make prefix=$HOME/local install
Install gflags:
版本:1.7,需要先安装gflags,再安装glog,其它依赖库可以并行安装
$ cd ~/temp/
$ git clone https://github.com/gflags/gflags/release/
$ mkdir build && cd build
$ bash
$ export CXXFLAGS="-fPIC"
$ ./configure --prefix=$HOME/local ..
$ make -j
$ make install
版本:2.1.1
$ cd ~/temp/
$ git clone https://github.com/gflags/gflags/release/
$ mkdir build && cd build
$ bash
$ CXXFLAGS="-fPIC" cmake -D CMAKE_INSTALL_PREFIX=$HOME/local ..
$ make -j
$ make install
版本2.1.2后需要较高版本的CMake。
Install glog:
版本:0.3.3
$ cd ~/temp/
$ wget https://google-glog.googlecode.com/files/glog-0.3.3.tar.gz
$ tar zxvf glog-0.3.3.tar.gz
$ cd glog-0.3.3
$ ./configure --prefix=$HOME/local
$ make && make install
Install hdf5:
matlab会自带hdf5,最好和matlab的版本号一致)。如果不一致,可以在linux设置环境变量
export HDF5_DISABLE_VERSION_CHECK=1
最新版版本号需要https://www.hdfgroup.org/ftp/HDF5/current/src/ 确认
$ cd ~/temp
$ wget "https://www.hdfgroup.org/ftp/HDF5/current/src/hdf5-1.8.14.tar"
$ tar -xf hdf5-1.8.14.tar
$ cd hdf5-1.8.14
$ ./configure --prefix=$HOME/local
$ make
$ make check # run test suite.
$ make install
$ make check-install # verify installation.
OpenBlas:
$ cd ~/temp/
$ git clone git://github.com/xianyi/OpenBLAS
$ cd OpenBlas
$ make FC=gfortran
$ make PREFIX=$HOME/local install