Ubuntu16.04 Caffe安装
2016-10-28 22:38
309 查看
环境:Ubuntu 16.04 64bit
Nvidia GeForce GTX 1070 (驱动Nvidia-Linux-x86_64-367.57)
CUDA 8.0.44
cuDNN 5.1
openCV 3.1.0
Nvidia驱动安装:Ubuntu16.04 Nvidia 显卡驱动安装
openCV安装:Ubuntu16.04 openCV3.1安装
Cuda、cuDNN安装:Ubuntu 16.04 CUDA 8 cuDNN 5.1安装
安装依赖
sudo apt-get update
sudo apt-get install uild-essential cmake git pkg-config
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install --no-install-recommends libboost-all-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev
sudo apt-get install python-pip python-dev python-numpy python-scipy # (Python general)
sudo apt-get install libblas-dev liblapack-dev libatlas-base-dev gfortran Python-numpy #(Python library)
sudo apt-get install libopencv-dev # (OpenCV 2.4)
sudo apt-get install git
Python接口依赖
其实上面包括了一些
sudo apt-get install protobuf-c-compiler protobuf-compiler
sudo apt-get install python-numpy python-scipy python-matplotlib python-sklearn python-skimage python-h5py python-protobuf python-leveldb python-networkx python-nose python-pandas python-gflags Cython ipython
安装BLAS
BLAS(基础线性代数集合)是一个应用程序接口标准。caffe推荐了三种实现:ATLAS, MKL, 或OpenBLAS。这里选择ATLAS。Intel mkl库的安装方法在底部参考中有。
sudo apt-get install libatlas-base-dev
下载caffe
git clone https://github.com/BVLC/caffe.git
修改Makefile
cd caffe
cp Makefile.config.example Makefile.config
gedit Makefile.config 或
vim Makefile.config
若不使用GPU,取消注释 #CPU_ONLY:=1
若使用cuDNN,取消注释 #USE_CUDNN:=1
若使用opencv3,取消注释 OPENCV_VERSION:=3
关于mkl库、python目录、anoconda目录、matlab目录、用Python编写Layer等的设置,我没有用到,在参考中有。
由于Ubuntu 16.04文件位置有些变化,将
# Whatever else you find you need goes here.
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib
修改为
# Whatever else you find you need goes here.
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
由于architecture不同hdf5目录可能不同,具体可以去目录里看看。比如x86_64可能是i386。具体找到hd5.h和libhdf5的目录,我还搭过一台ubuntu 16.04.1的服务器,include和library目录(hdf5.h和libhdf5的目录)分别是/usr/include/和/usr/lib/x86_64-linux-gnu/
编译
make all
make test
make runtest
编译时可根据电脑配置加上-j4或-j8或-16进行多核运算加速编译,但数值设的太大可能会出错
Python接口:
make pycaffe
然后在环境变量PYTHONPATH里加上/path/to/caffe/python,在python终端里import caffe看看是否可行。
参考:
Ubuntu 16.04上安装Caffe
Ubuntu 16.04下Matlab2014a+Anaconda2+OpenCV3.1+Caffe安装
Ubuntu16.04+CUDA8.0+caffe配置
Ubuntu16.04 + cuda8.0 + GTX1080 + Opencv3.0 + caffe 安装教程
Caffe学习:pycaffe接口配置
Nvidia GeForce GTX 1070 (驱动Nvidia-Linux-x86_64-367.57)
CUDA 8.0.44
cuDNN 5.1
openCV 3.1.0
Nvidia驱动安装:Ubuntu16.04 Nvidia 显卡驱动安装
openCV安装:Ubuntu16.04 openCV3.1安装
Cuda、cuDNN安装:Ubuntu 16.04 CUDA 8 cuDNN 5.1安装
安装依赖
sudo apt-get update
sudo apt-get install uild-essential cmake git pkg-config
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install --no-install-recommends libboost-all-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev
sudo apt-get install python-pip python-dev python-numpy python-scipy # (Python general)
sudo apt-get install libblas-dev liblapack-dev libatlas-base-dev gfortran Python-numpy #(Python library)
sudo apt-get install libopencv-dev # (OpenCV 2.4)
sudo apt-get install git
Python接口依赖
其实上面包括了一些
sudo apt-get install protobuf-c-compiler protobuf-compiler
sudo apt-get install python-numpy python-scipy python-matplotlib python-sklearn python-skimage python-h5py python-protobuf python-leveldb python-networkx python-nose python-pandas python-gflags Cython ipython
安装BLAS
BLAS(基础线性代数集合)是一个应用程序接口标准。caffe推荐了三种实现:ATLAS, MKL, 或OpenBLAS。这里选择ATLAS。Intel mkl库的安装方法在底部参考中有。
sudo apt-get install libatlas-base-dev
下载caffe
git clone https://github.com/BVLC/caffe.git
修改Makefile
cd caffe
cp Makefile.config.example Makefile.config
gedit Makefile.config 或
vim Makefile.config
若不使用GPU,取消注释 #CPU_ONLY:=1
若使用cuDNN,取消注释 #USE_CUDNN:=1
若使用opencv3,取消注释 OPENCV_VERSION:=3
关于mkl库、python目录、anoconda目录、matlab目录、用Python编写Layer等的设置,我没有用到,在参考中有。
由于Ubuntu 16.04文件位置有些变化,将
# Whatever else you find you need goes here.
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib
修改为
# Whatever else you find you need goes here.
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
由于architecture不同hdf5目录可能不同,具体可以去目录里看看。比如x86_64可能是i386。具体找到hd5.h和libhdf5的目录,我还搭过一台ubuntu 16.04.1的服务器,include和library目录(hdf5.h和libhdf5的目录)分别是/usr/include/和/usr/lib/x86_64-linux-gnu/
编译
make all
make test
make runtest
编译时可根据电脑配置加上-j4或-j8或-16进行多核运算加速编译,但数值设的太大可能会出错
Python接口:
make pycaffe
然后在环境变量PYTHONPATH里加上/path/to/caffe/python,在python终端里import caffe看看是否可行。
参考:
Ubuntu 16.04上安装Caffe
Ubuntu 16.04下Matlab2014a+Anaconda2+OpenCV3.1+Caffe安装
Ubuntu16.04+CUDA8.0+caffe配置
Ubuntu16.04 + cuda8.0 + GTX1080 + Opencv3.0 + caffe 安装教程
Caffe学习:pycaffe接口配置
相关文章推荐
- Ubuntu16.04下安装Caffe
- Ubuntu16.04安装配置GPU版本Caffe经验总结
- ubuntu16.04安装caffe以及各种问题汇总
- ubuntu16.04 64位 cpu安装tensorflow+theano+keras+caffe+xgboost
- Ubuntu16.04 Caffe安装笔记
- 虚拟机ubuntu16.04 安装caffe
- ubuntu16.04 安装 caffe cuda 相关流程
- Ubuntu16.04 Caffe 安装步骤记录(超详尽)
- 【caffe】caffe安装 ubuntu16.04 版
- No module named caffe,ubuntu16.04安装caffe的问题解决
- Ubuntu16.04+cuda8.0+caffe安装教程
- Caffe安装教程之Ubuntu16.04--make all报错
- Caffe+Ubuntu 16.04 安装教程
- Ubuntu16.04下安装Cuda8.0+Caffe+TensorFlow-gpu+Pycharm过程(Simple)
- Ubuntu16.04中caffe安装(only cpu)
- Ubuntu14.04/16.04安装caffe
- Ubuntu16.04 Caffe 安装步骤记录(超详尽)
- caffe安装(1)ubuntu16.04+显卡驱动+cuda8.0
- ubuntu16.04+gtx1060+cuda8.0+caffe安装、测试经历
- Ubuntu 16.04 Cuda 8.0 Opencv 3.1.0 Anaconda2 Caffe 安装