设置ubuntu的softwares&updates的源为国内源,这样会提高下载速度。
如果是安装python相关库,为提高速度使用:
pip3 install 要下载的库 -i https://国内源
安装caffe依赖库:
1 # python3 modules (numpy, protobuf, skimage) 2 sudo pip3 install numpy 3 sudo apt-get install python3-skimage 4 sudo apt-get install python3-protobuf 5 6 # build essential 7 sudo apt-get install build-essential cmake git pkg-config 8 9 # gflags, glog, lmdb 10 sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev 11 12 # boost 13 sudo apt-get install libboost-all-dev 14 15 # hdf5 16 sudo apt-get install libhdf5-dev 17 18 # protobuf 19 sudo apt-get install protobuf-compiler libprotobuf-dev 20 21 # blas 22 sudo apt-get install libblas-dev libcblas-dev libatlas-base-dev libopenblas-dev 23 24 # leveldb 25 sudo apt-get install libleveldb-dev 26 27 # snappy 28 sudo apt-get install libsnappy-dev
安装opencv3.2
下载源代码 opencv-3.2.0.zip
1 unzip opencv-3.2.0.zip 2 cd opencv-3.2.0 3 mkdir release 4 cd release 5 cmake .. 6 make -j4 7 sudo make install
安装git:
sudo apt-get install git
安装caffe:
git clone https://github.com/BVLC/caffe
cd caffe
cp Makefile.config.example Makefile.config
Makefile.config文件:
1 ## Refer to http://caffe.berkeleyvision.org/installation.html 2 # Contributions simplifying and improving our build system are welcome! 3 4 # cuDNN acceleration switch (uncomment to build with cuDNN). 5 # USE_CUDNN := 1 6 7 # CPU-only switch (uncomment to build without GPU support). 8 CPU_ONLY := 1 9 10 # uncomment to disable IO dependencies and corresponding data layers 11 # USE_OPENCV := 0 12 # USE_LEVELDB := 0 13 # USE_LMDB := 0 14 # This code is taken from https://github.com/sh1r0/caffe-android-lib 15 # USE_HDF5 := 0 16 17 # uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary) 18 # You should not set this flag if you will be reading LMDBs with any 19 # possibility of simultaneous read and write 20 # ALLOW_LMDB_NOLOCK := 1 21 22 # Uncomment if you're using OpenCV 3 23 OPENCV_VERSION := 3 24 25 # To customize your choice of compiler, uncomment and set the following. 26 # N.B. the default for Linux is g++ and the default for OSX is clang++ 27 # CUSTOM_CXX := g++ 28 29 # CUDA directory contains bin/ and lib/ directories that we need. 30 CUDA_DIR := /usr/local/cuda 31 # On Ubuntu 14.04, if cuda tools are installed via 32 # "sudo apt-get install nvidia-cuda-toolkit" then use this instead: 33 # CUDA_DIR := /usr 34 35 # CUDA architecture setting: going with all of them. 36 # For CUDA < 6.0, comment the *_50 through *_61 lines for compatibility. 37 # For CUDA < 8.0, comment the *_60 and *_61 lines for compatibility. 38 # For CUDA >= 9.0, comment the *_20 and *_21 lines for compatibility. 39 CUDA_ARCH := -gencode arch=compute_20,code=sm_20 \ 40 -gencode arch=compute_20,code=sm_21 \ 41 -gencode arch=compute_30,code=sm_30 \ 42 -gencode arch=compute_35,code=sm_35 \ 43 -gencode arch=compute_50,code=sm_50 \ 44 -gencode arch=compute_52,code=sm_52 \ 45 -gencode arch=compute_60,code=sm_60 \ 46 -gencode arch=compute_61,code=sm_61 \ 47 -gencode arch=compute_61,code=compute_61 48 49 # BLAS choice: 50 # atlas for ATLAS (default) 51 # mkl for MKL 52 # open for OpenBlas 53 BLAS := atlas 54 # Custom (MKL/ATLAS/OpenBLAS) include and lib directories. 55 # Leave commented to accept the defaults for your choice of BLAS 56 # (which should work)! 57 # BLAS_INCLUDE := /path/to/your/blas 58 # BLAS_LIB := /path/to/your/blas 59 60 # Homebrew puts openblas in a directory that is not on the standard search path 61 # BLAS_INCLUDE := $(shell brew --prefix openblas)/include 62 # BLAS_LIB := $(shell brew --prefix openblas)/lib 63 64 # This is required only if you will compile the matlab interface. 65 # MATLAB directory should contain the mex binary in /bin. 66 # MATLAB_DIR := /usr/local 67 # MATLAB_DIR := /Applications/MATLAB_R2012b.app 68 69 # NOTE: this is required only if you will compile the python interface. 70 # We need to be able to find Python.h and numpy/arrayobject.h. 71 #PYTHON_INCLUDE := /usr/include/python2.7 \ 72 # /usr/lib/python2.7/dist-packages/numpy/core/include 73 # Anaconda Python distribution is quite popular. Include path: 74 # Verify anaconda location, sometimes it's in root. 75 # ANACONDA_HOME := $(HOME)/anaconda 76 # PYTHON_INCLUDE := $(ANACONDA_HOME)/include \ 77 # $(ANACONDA_HOME)/include/python2.7 \ 78 # $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include 79 80 # Uncomment to use Python 3 (default is Python 2) 81 PYTHON_LIBRARIES := boost_python3 python3.6m 82 PYTHON_INCLUDE := /usr/include/python3.6m \ 83 # /usr/lib/python3.6/dist-packages/numpy/core/include 84 85 # We need to be able to find libpythonX.X.so or .dylib. 86 PYTHON_LIB := /usr/lib 87 # PYTHON_LIB := $(ANACONDA_HOME)/lib 88 89 # Homebrew installs numpy in a non standard path (keg only) 90 # PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include 91 # PYTHON_LIB += $(shell brew --prefix numpy)/lib 92 93 # Uncomment to support layers written in Python (will link against Python libs) 94 WITH_PYTHON_LAYER := 1 95 96 # Whatever else you find you need goes here. 97 INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial 98 LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial 99 100 # If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies 101 # INCLUDE_DIRS += $(shell brew --prefix)/include 102 # LIBRARY_DIRS += $(shell brew --prefix)/lib 103 104 # NCCL acceleration switch (uncomment to build with NCCL) 105 # https://github.com/NVIDIA/nccl (last tested version: v1.2.3-1+cuda8.0) 106 # USE_NCCL := 1 107 108 # Uncomment to use `pkg-config` to specify OpenCV library paths. 109 # (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.) 110 # USE_PKG_CONFIG := 1 111 112 # N.B. both build and distribute dirs are cleared on `make clean` 113 BUILD_DIR := build 114 DISTRIBUTE_DIR := distribute 115 116 # Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171 117 # DEBUG := 1 118 119 # The ID of the GPU that 'make runtest' will use to run unit tests. 120 TEST_GPUID := 0 121 122 # enable pretty build (comment to see full commands) 123 Q ?= @
View Code
本站文章如无特殊说明,均为本站原创,如若转载,请注明出处:Ubuntu18.04安装caffe python3.6 opencv3.2 CPU - Python技术站