caffe model

https://github.com/soeaver/caffe-model/tree/master/cls

Install caffe on Ubuntu 16.04 with gpu

useful links:

https://github.com/BVLC/caffe/wiki/Ubuntu-16.04-or-15.10-Installation-Guide https://blog.csdn.net/wuzuyu365/article/details/52430657

Install thirdparty packages

1. install packages

 
	sudo apt update
	sudo apt upgrade
	sudo apt install -y build-essential cmake git pkg-config
	sudo apt install -y libprotobuf-dev libleveldb-dev libsnappy-dev libhdf5-serial-dev protobuf-compiler
	sudo apt install -y libatlas-base-dev 
	sudo apt install -y --no-install-recommends libboost-all-dev
	sudo apt install -y libgflags-dev libgoogle-glog-dev liblmdb-dev

2. install python

3. install cuda (9.0), cudnn (7.0), opencv (3.4.0)

3.1 安裝驅動

主要步骤：

1.卸载系统里的Nvidia低版本显卡驱动

	sudo apt-get purge nvidia*

2.把显卡驱动加入PPA

	sudo add-apt-repository ppa:graphics-drivers
	sudo apt-get update

3.查找安裝显卡驱动最新的版本号
查找并安装最新驱动
	sudo apt-cache search nvidia
或可使用终端命令查看Ubuntu推荐的驱动版本：
	ubuntu-drivers devices
采用apt-get命令在终端安装：
	sudo apt-get install nvidia-430 nvidia-settings nvidia-prime
4.重啓驗證
	sudo reboot
	nvidia-smi

3.2 install cuda (9.0)

https://blog.csdn.net/Angela_happy/article/details/80977265

error
<pre>
	Missing recommended library: libGLU.so
	Missing recommended library: libXi.so
	Missing recommended library: libXmu.so
</pre>
solve:
<pre>
	sudo apt-get install libglu1-mesa libxi-dev libxmu-dev libglu1-mesa-dev
</pre>

3.3 install cudnn (7.0)

cudnn的安装非常简单

https://blog.csdn.net/wanzhen4330/article/details/81699769

（1）下载安装文件

按需求下载cudnn的安装文件：

https://developer.nvidia.com/rdp/cudnn-archive

（2）安装cudnn

解压下载的文件，可以看到cuda文件夹，在当前目录打开终端，执行如下命令：

	tar -zxvf cudnn-8.0-linux-x64-v6.0.tgz 

	sudo cp cuda/include/cudnn.h /usr/local/cuda/include/

	sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64/ -d

	sudo chmod a+r /usr/local/cuda/include/cudnn.h

	sudo chmod a+r /usr/local/cuda/lib64/libcudnn*
	
（3）查看cudnn版本
 
在终端输入

	cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

如果出现以下信息，说明安装成功。 

	#define CUDNN_MAJOR 7
	#define CUDNN_MINOR 0
	#define CUDNN_PATCHLEVEL 5
		--
	#define CUDNN_VERSION    (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

	#include "driver_types.h"

安装完成后可用 nvcc -V 命令验证是否安装成功，若出现以下信息则表示安装成功：

	yhao@yhao-X550VB:~$ nvcc -V
	nvcc: NVIDIA (R) Cuda compiler driver
	Copyright (c) 2005-2016 NVIDIA Corporation
	Built on Tue_Jan_10_13:22:03_CST_2017
	Cuda compilation tools, release 8.0, V8.0.61

3.4 install opencv (3.4.0)

https://blog.csdn.net/xd1723138323/article/details/80498381

Adapt caffe config files

download caffe: github link
If download caffe from github link, the version should be 1.0. Go to caffe folder, edit /src/caffe/util/blocking_queue.cpp, after line 89, add new line: template class BlockingQueue<Datum*>;
Modify Makefile.config

3.1 cp Makefile.config.example Makefile.config
3.2 edit Makefile.config, find and modify the following lines:

	PYTHON_INCLUDE := /usr/include/python2.7 /usr/local/lib/python2.7/dist-packages/numpy/core/include  
	WITH_PYTHON_LAYER := 1  
	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 /usr/lib/x86_64-linux-gnu/hdf5/serial

3.3 since cuda 8.0, compute capability 2.0 and 2.1 are discarded, so delete the following two lines:

	-gencode arch=compute_20,code=sm_20
	-gencode arch=compute_20,code=sm_21

fix libhdf5 links

	find . -type f -exec sed -i -e 's^"hdf5.h"^"hdf5/serial/hdf5.h"^g' -e 's^"hdf5_hl.h"^"hdf5/serial/hdf5_hl.h"^g' '{}' \;
	cd /usr/lib/x86_64-linux-gnu
	sudo ln -s libhdf5_serial.so.10.1.0 libhdf5.so
	sudo ln -s libhdf5_serial_hl.so.10.0.2 libhdf5_hl.so

install python packages

	cd python
	for req in $(cat requirements.txt); do sudo -H pip install $req --upgrade; done

Modify Makefile replace this line:

	NVCCFLAGS += -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)

with the following line:

	NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)

Modify CMakeLists.txt add the following line:

	# ---[ Includes
	set(${CMAKE_CXX_FLAGS} "-D_FORCE_INLINES ${CMAKE_CXX_FLAGS}")

Build and Compile caffe

	make all -j8
	make test -j8
	sudo make runtest -j8
	sudo make pycaffe -j8 (add "export PYTHONPATH=/home/caffe/python:$PYTHONPATH" to "~/.bashrc")

Debug

error at make all:

	caffe.cpp:(.text+0x15eb): undefined reference to `caffe::Net::CopyTrainedLayersFrom(std::__cxx11::basic_string, std::allocator > const&)` collect2: error: ld returned 1 exit status

solve:

	delete libcaffe.so present under /usr/local/lib

links: rbgirshick/py-faster-rcnn#477 BVLC/caffe#3396

error:

	.build_release/tools/caffe: error while loading shared libraries: libcudart.so.9.0: cannot open shared object file: No such file or directory

solve:

	in ~/.bashrc, should have the two lines:
		export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
		export PATH=/usr/local/cuda/bin:$PATH
	sudo ldconfig /usr/local/cuda/lib64

links: BVLC/caffe#4944

error at make pycaffe:

	python/caffe/_caffe.cpp:10:31: fatal error: numpy/arrayobject.h: No such file or directory

solve:

	sudo apt install python-numpy

links: https://blog.csdn.net/wuzuyu365/article/details/52430657

error at import caffe in python:

	ImportError: No module named skimage.io

solve:

	pip install scikit-image
	sudo apt-get install python-matplotlib python-numpy python-pil python-scipy
	sudo apt-get install build-essential cython
	sudo apt-get install python-skimage

links: yahoo/open_nsfw#13 BVLC/caffe#50

pip错误 ImportError: No module named _internal

error at import caffe in python:

	ImportError: No module named google.protobuf.internal

solve:

	sudo apt install python-protobuf

links: https://stackoverflow.com/questions/37666241/importing-caffe-results-in-importerror-no-module-named-google-protobuf-interna https://stackoverflow.com/questions/37666241/importing-caffe-results-in-importerror-no-module-named-google-protobuf-interna/37905483

error pip错误 ImportError: No module named _internal

	Traceback (most recent call last):
	File "/home/ubuntu/.local/bin/pip", line 7, in 
	 
	from pip._internal import main
	 
	ImportError: No module named _internal

solve:

	强制重新安装pip3
	wget https://bootstrap.pypa.io/get-pip.py  --no-check-certificate
	 
	sudo python3 get-pip.py --force-reinstall

	安装 requirements.txt列表
	pip install -r requirements.txt -i --user
	
	
	通过cd指令转到caffe下的python目录中运行下面指令

	for req in $(cat requirements.txt); do pip install $req; done

	如果提示权限不够的话就是

	for req in $(cat requirements.txt); do sudo pip install $req; done

error：

	from pip import __main__
	ImportError: No module named 'pip'

solve:

	wget https://bootstrap.pypa.io/get-pip.py  --no-check-certificate
	sudo python get-pip.py

error：

	g++: internal compiler error: Killed (program cc1plus)
	Please submit a full bug report,

solve:

	主要原因大体上是因为内存不足,有点坑 临时使用交换分区来解决吧

 

	sudo dd if=/dev/zero of=/swapfile bs=64M count=16
	sudo mkswap /swapfile
	sudo swapon /swapfile

	After compiling, you may wish to


	Code:
	sudo swapoff /swapfile
	sudo rm /swapfile

install python3.5

【说明】

  正常情况下你的ubuntu系统是已经自带了python的，不过自带的版本是2.7的，而现在的Python3.5和2.7其实已经非常不同了，作为开发学习的话还是新版本的Python3.5吧。

【安装】

  1、首先python不在ubuntu的软件仓库，所以我们需要去PPA上找软件源，打开终端，输入下面的命令：sudo add-apt-repository ppa:fkrull/deadsnakes  

  2、fkrull/deadsnakes就是ubuntu提供的python的repository。接下来接着输入：sudo apt-get update  

  3、sudo apt-get install python3.5

【修改配置】

  1、安装完成之后，你在终端中输入python，输出的信息里面会提示是2.7版本的，也就是说默认打开的并不是刚才安装好的3.5，所以还需要我们重新修改一下链接。方法如下：第一步：先备份原来的链接（在对系统文件执行删除之前进行备份是个好习惯）

  sudo cp /usr/bin/python /usr/bin/python_bak
  2、第二步：删除原来的指向2.7版本的默认链接：sudo rm /usr/bin/python  

  3、第三步：重新指定链接指向3.5版本：sudo ln -s /usr/bin/python3.5 /usr/bin/python  

  4、大功告成，此时在终端再输入python，输出的信息就是3.5版本了。

caffe+python3.5的配置

	打开Makefile.config
	去掉下面的注释：

	PYTHON_LIBRARIES := boost_python3 python3.5m
	PYTHON_INCLUDE := /usr/include/python3.5m \
                                            /usr/lib/python3.5/dist-packages/numpy/core/include

	然后重新编译。

	如果编译过程中遇到cannot find -lboost_python3     

	查看/usr/lib/x86_64-linux-gnu下只有：libboost_python-py35.so

 	问题解决办法，将boost_python3改成boost_python-py35

	PYTHON_LIBRARIES := boost_python-py35 python3.5m

https://github.com/guonannan123456/restnet50/blob/master/caffe%20installation.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

caffe installation.md

caffe installation.md

caffe model

Install caffe on Ubuntu 16.04 with gpu

useful links:

Install thirdparty packages

1. install packages

2. install python

3. install cuda (9.0), cudnn (7.0), opencv (3.4.0)

Adapt caffe config files

Build and Compile caffe

Debug

Files

caffe installation.md

Latest commit

History

caffe installation.md

File metadata and controls

caffe model

Install caffe on Ubuntu 16.04 with gpu

useful links:

Install thirdparty packages

1. install packages

2. install python

3. install cuda (9.0), cudnn (7.0), opencv (3.4.0)

Adapt caffe config files

Build and Compile caffe

Debug