ubuntu系統下配置caffe的GPU環境，訓練mobileNet-SSD並在EAIDK-310嵌入式上推理

前言
一、NVIDIA顯示卡驅動及KUDA的安裝
- 1.安裝gpu驅動
- 2.安裝cuda和cudnn
二、ubuntu安裝GPU版caffe
- (一) 安裝依賴庫:
- (二) 下載原始碼並切換到 ssd 分支
- (三) 修改配置檔案
- - 1.修改Makefile.config檔案
  - 2.修改Makefile檔案
- (四) 編譯及測試caffe
三、anaconda3環境下安裝gpu版caffe
- (一) ubuntu系統安裝anaconda3
- - 1.下載及安裝
  - 2.建立虛擬環境
- (二) 虛擬環境下安裝caffe
四、caffe環境下訓練ＭobileNet—SSD

前言

如果沒有安裝好ubuntu系統或者系統崩潰需要重灌的可以參考上一篇部落格：ubuntu+win10雙系統安裝。

一、NVIDIA顯示卡驅動及KUDA的安裝

1.安裝gpu驅動

這裡推薦一個簡單的安裝方法，開啟選單，找到軟體和更新，
找到附加驅動，這裡會自動推薦適合GPU的驅動，
選擇使用NVIDIA驅動，點選右下角的應用更改，系統會自動安裝顯示卡驅動並重啟。
若此方法失敗，則可參考網上安裝gpu驅動的教程，教程有很多，這裡不再贅述。
驗證是否成功安裝：nvidia-smi

nvidia-smi

顯示如下圖則說明安裝成功，注意版本，我這裡是Driver Version: 384.130

2.安裝cuda和cudnn

版本問題
cada: cuda_8.0.44_linux.run
cudnn:cudnn-8.0-linux-x64-v5.1.tgz

如沒有使用此版本以及上面的顯示卡驅動版本，不保證能配置成功。本人經過長達兩週的時間摸索此對應的版本。推薦CUDA官網，cudnn的安裝下載，若有需要，可參考百度雲。
至於安裝教程，網上很多，這裡推薦一篇教程。
檢視是否安裝成功：nvcc -V

二、ubuntu安裝GPU版caffe

(一) 安裝依賴庫:

1.安裝指令如下：

sudo apt-get install libprotobuf-dev //（16.04預設版本2.6.1，不能升級）sudo apt-get install libleveldb-devsudo apt-get install libsnappy-devsudo apt-get install libopencv-devsudo apt-get install libhdf5-serial-devsudo apt-get install protobuf-compilersudo apt-get install --no-install-recommends libboost-all-devsudo apt-get install libatlas-base-devsudo apt-get install python-devsudo apt-get install libgflags-devsudo apt-get install libgoogle-glog-devsudo apt-get install liblmdb-dev

2.特別注意
本人安裝時python版本是Ubuntu16.04自帶的Python2.7，如果需要其他python版本，建議參考後面第三節anaconda虛擬環境下的安裝。
安裝完成後確認python版本,直接輸入python：

python

這裡我只有用opencv2.4.9.1才編譯成功，及Ubuntu16.04預設安裝的版本。
安裝完成後確認opencv版本,接著輸入：

import cv2cv2.__version__

(二) 下載原始碼並切換到 ssd 分支

1.下載碼源
在主目錄下(我的是/home/xxx,xxx為使用者名稱)開啟命令視窗，依次輸入：
下載github上的caffe專案：

git clone https://github.com/weiliu89/caffe.git

若後面編譯不成功，也可以下載這裡編譯好的caffe。
切換到caffe資料夾：

cd caffe

2.切換到 ssd 分支

git checkout ssd

3.安裝python依賴庫
切換到 caffe 下的 python 目錄下，下載 python 依賴庫(先安裝 pip)

cd pythonsudo apt-get install python-pipfor req in $(catpython requirements.txt); do pip install $req; done

(三) 修改配置檔案

（針對python2.7,opencv2.4.9.1d的環境修改）

1.修改Makefile.config檔案

先將Makefile.config.example複製並重命名為Makefile.config

複製指令如下：

cp Makefile.config.example Makefile.config

然後修改Makefile.config檔案，修改後如下：

## Refer to http://caffe.berkeleyvision.org/installation.html# Contributions simplifying and improving our build system are welcome!# cuDNN acceleration switch (uncomment to build with cuDNN).USE_CUDNN := 1# CPU-only switch (uncomment to build without GPU support).# CPU_ONLY := 1# uncomment to disable IO dependencies and corresponding data layers# USE_OPENCV := 0# USE_LEVELDB := 0# USE_LMDB := 0# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)#	You should not set this flag if you will be reading LMDBs with any#	possibility of simultaneous read and write# ALLOW_LMDB_NOLOCK := 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++# CUDA directory contains bin/ and lib/ directories that we need.CUDA_DIR := /usr/local/cuda# On Ubuntu 14.04, if cuda tools are installed via# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:# CUDA_DIR := /usr# CUDA architecture setting: going with all of them.# For CUDA < 6.0, comment the lines after *_35 for compatibility.CUDA_ARCH := -gencode arch=compute_30,code=sm_30 \             -gencode arch=compute_35,code=sm_35 \             -gencode arch=compute_50,code=sm_50 \             -gencode arch=compute_52,code=sm_52 \             -gencode arch=compute_61,code=sm_61# BLAS choice:# atlas for ATLAS (default)# mkl for MKL# open for OpenBlas# BLAS := atlasBLAS := open# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.# Leave commented to accept the defaults for your choice of BLAS# (which should work)!# BLAS_INCLUDE := /path/to/your/blas# BLAS_LIB := /path/to/your/blas# Homebrew puts openblas in a directory that is not on the standard search path# BLAS_INCLUDE := $(shell brew --prefix openblas)/include# BLAS_LIB := $(shell brew --prefix openblas)/lib# This is required only if you will compile the matlab interface.# MATLAB directory should contain the mex binary in /bin.# MATLAB_DIR := /usr/local# MATLAB_DIR := /Applications/MATLAB_R2012b.app# 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/local/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)/anaconda2# PYTHON_INCLUDE := $(ANACONDA_HOME)/include \		$(ANACONDA_HOME)/include/python2.7 \		$(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \# Uncomment to use Python 3 (default is Python 2)# PYTHON_LIBRARIES := boost_python3 python3.5m# PYTHON_INCLUDE := /usr/include/python3.5m \#                 /usr/lib/python3.5/dist-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# Homebrew installs numpy in a non standard path (keg only)# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include# PYTHON_LIB += $(shell brew --prefix numpy)/lib# Uncomment to support layers written in Python (will link against Python libs)# WITH_PYTHON_LAYER := 1# Whatever else you find you need goes here.INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include \		/usr/include/hdf5/serialLIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib \		/usr/lib/x86_64-linux-gnu/hdf5/serial# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies# INCLUDE_DIRS += $(shell brew --prefix)/include# LIBRARY_DIRS += $(shell brew --prefix)/lib# Uncomment to use `pkg-config` to specify OpenCV library paths.# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)# USE_PKG_CONFIG := 1# N.B. both build and distribute dirs are cleared on `make clean`BUILD_DIR := buildDISTRIBUTE_DIR := distribute# Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171# DEBUG := 1# The ID of the GPU that 'make runtest' will use to run unit tests.TEST_GPUID := 0# enable pretty build (comment to see full commands)Q ?= @

2.修改Makefile檔案

（這裡程式碼有點多，避免一行行的去查詢修改，這裡直接給出修改好的Makefile檔案。）

(四) 編譯及測試caffe

make pycaffe -j8make all -j8make test -j8make runtest -j8

時間較長，請耐心等待，-j8是指多核同時編譯，可節省時間，根據自己電腦設定-j**。出現pass則編譯成功。

測試：
新增環境變數，注意caffe的路徑：

cd caffe/pythonexport PYTHONPATH=~/caffe/python:$PYTHONPATH

進入python環境，輸入：import caffe如果沒有報錯，證明安裝成功。若編譯通過但是匯入失敗，則make pycaffe一下。

匯入環境變數：

sudo echo export PYTHONPATH=“~/caffe/python" >>~/.bashrcsource ~/.bashrc #使環境變數生效

或者：

echo "export CAFFE_ROOT=~/caffe" >> ~/.bashrcecho "export PYTHONPATH=${CAFFE_ROOT}/python:$PYTHONPATH" >> ~/.bashrcsource ~/.bashrc

如出錯參考連結。

重新開啟一個終端視窗，輸入python，再輸入caffe，不出錯則新增環境變數成功。

三、anaconda3環境下安裝gpu版caffe

(一) ubuntu系統安裝anaconda3

（因為這裡主要介紹anaconda3下的caffe環境配置，至於anaconda3的安裝，不做過多的贅述。）

1.下載及安裝

下載：anaconda安裝包下載，推薦官網，或者百度雲，提取碼：zu0i。

安裝：網上很多教程，這裡推薦一篇。

是否安裝成功：安裝完成後需要新增環境變數，檢視環境變數終端輸入：

sudo gedit ~/.bashrc

開啟.bashrc後看最後是否有如下字樣程式碼

## >>> conda initialize >>>## !! Contents within this block are managed by 'conda init' !!__conda_setup="$('/home/×××/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"if [ $? -eq 0 ]; then    eval "$__conda_setup"else    if [ -f "/home/×××/anaconda3/etc/profile.d/conda.sh" ]; then        . "/home/×××/anaconda3/etc/profile.d/conda.sh"    else       export PATH="/home/×××/anaconda3/bin:$PATH"    fifiunset __conda_setup##    <<< conda initialize <<<

若沒有則複製進去，注意×××是自己的使用者名稱。
安裝完成後，終端命令列前出現預設環境(base)，輸入conda不報錯說明安裝成功。

2.建立虛擬環境

建立名為caffe的python2.7的虛擬環境並切換至該環境：

conda create -n caffe python=2.7conda activate caffe

(二) 虛擬環境下安裝caffe

安裝依賴檔案如第二節所示，但是我遇到的問題是sudo apt-get安裝的opencv庫在conda虛擬環境下無法呼叫，重新編譯安裝的opencv庫與我想安裝的caffe環境又有衝突，導致caffe編譯失敗。
這裡給出我的解決方法：
將usr/lib/python2.7/dist-packages目錄下的cv.py,cv.pyc和cv2.x86_64-linux-gnu.so三個檔案複製到anaconda3/envs/caffe/lib/python2.7/dist-packages資料夾下，並將cv2.x86_64-linux-gnu.so改名為cv2.so

至此，caffe下的cv2版本才符合要求，如圖所示：
下載碼源（與前面一致）
在主目錄下(我的是/home/xxx/my_code,xxx為使用者名稱)開啟命令視窗，依次輸入：
下載github上的caffe專案：
```
git clone https://github.com/weiliu89/caffe.git
```
切換到caffe資料夾：
```
cd caffe
```
切換到 ssd 分支

git checkout ssd

安裝python依賴庫
切換到 caffe 下的 python 目錄下，下載 python 依賴庫(先安裝 pip)

cd pythonsudo apt-get install python-pipfor req in $(catpython requirements.txt); do pip install $req; done

修改配置檔案
首先是Makefile.config檔案，將python的路徑改成anaconda下虛擬環境的python，其他的不變。
修改後如下：

## Refer to http://caffe.berkeleyvision.org/installation.html# Contributions simplifying and improving our build system are welcome!# cuDNN acceleration switch (uncomment to build with cuDNN).USE_CUDNN := 1# CPU-only switch (uncomment to build without GPU support).# CPU_ONLY := 1# uncomment to disable IO dependencies and corresponding data layers# USE_OPENCV := 0# USE_LEVELDB := 0# USE_LMDB := 0# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)#	You should not set this flag if you will be reading LMDBs with any#	possibility of simultaneous read and write# ALLOW_LMDB_NOLOCK := 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++# CUDA directory contains bin/ and lib/ directories that we need.CUDA_DIR := /usr/local/cuda# On Ubuntu 14.04, if cuda tools are installed via# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:# CUDA_DIR := /usr# CUDA architecture setting: going with all of them.# For CUDA < 6.0, comment the lines after *_35 for compatibility.CUDA_ARCH := -gencode arch=compute_30,code=sm_30 \             -gencode arch=compute_35,code=sm_35 \             -gencode arch=compute_50,code=sm_50 \             -gencode arch=compute_52,code=sm_52 \             -gencode arch=compute_61,code=sm_61# BLAS choice:# atlas for ATLAS (default)# mkl for MKL# open for OpenBlas# BLAS := atlasBLAS := open# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.# Leave commented to accept the defaults for your choice of BLAS# (which should work)!# BLAS_INCLUDE := /path/to/your/blas# BLAS_LIB := /path/to/your/blas# Homebrew puts openblas in a directory that is not on the standard search path# BLAS_INCLUDE := $(shell brew --prefix openblas)/include# BLAS_LIB := $(shell brew --prefix openblas)/lib# This is required only if you will compile the matlab interface.# MATLAB directory should contain the mex binary in /bin.# MATLAB_DIR := /usr/local# MATLAB_DIR := /Applications/MATLAB_R2012b.app# 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/local/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/lf/anaconda3/envs/caffePYTHON_INCLUDE := $(ANACONDA_HOME)/include \		$(ANACONDA_HOME)/include/python2.7 \		$(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \# Uncomment to use Python 3 (default is Python 2)# PYTHON_LIBRARIES := boost_python3 python3.5m# PYTHON_INCLUDE := /usr/include/python3.5m \#                 /usr/lib/python3.5/dist-packages/numpy/core/include# We need to be able to find libpythonX.X.so or .dylib.# PYTHON_LIB := /usr/libPYTHON_LIB := $(ANACONDA_HOME)/libLINKFLAGS := -Wl,-rpath,$(ANACONDA_HOME)/lib# Homebrew installs numpy in a non standard path (keg only)# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include# PYTHON_LIB += $(shell brew --prefix numpy)/lib# Uncomment to support layers written in Python (will link against Python libs)WITH_PYTHON_LAYER := 1# Whatever else you find you need goes here.INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include \		/usr/include/hdf5/serialLIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib \		/usr/lib/x86_64-linux-gnu/hdf5/serial# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies# INCLUDE_DIRS += $(shell brew --prefix)/include# LIBRARY_DIRS += $(shell brew --prefix)/lib# Uncomment to use `pkg-config` to specify OpenCV library paths.# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)# USE_PKG_CONFIG := 1# N.B. both build and distribute dirs are cleared on `make clean`BUILD_DIR := buildDISTRIBUTE_DIR := distribute# Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171# DEBUG := 1# The ID of the GPU that 'make runtest' will use to run unit tests.TEST_GPUID := 0# enable pretty build (comment to see full commands)Q ?= @

其次是Makefile檔案，程式碼比較多，避免一行行的去查詢修改，這裡直接給出修改好的Makefile檔案。）

剩下的編譯過程也是完全一樣，請參考前面部分。

至此，anconda下的caffe環境就安裝好了。

四、caffe環境下訓練ＭobileNet—SSD

本人安裝ubuntu下的caffe環境主要是利用自己的資料集訓練ＭobileNetSSD model，這部分可以參考連結。
最後，將訓練的模型.caffemodel模型檔案轉化成.tmfile檔案，在eaidk-310平臺上做推理，這部分內容可以參考下一篇部落格。

150-ubuntu16.04系統下配置caffe的GPU環境，訓練mobileNet

前言