ubuntu16.04下安装caffe以及错误汇总(转）

拼搏现实的明天。 2022-11-25 13:13 165阅读 0赞

这篇文章主要讲安装和编译caffe过程以及遇到的问题，还有解决方法。关于如何安装nvidia驱动，cuda，cudnn，以及opencv3 等网上已经有足够多的教程了，可以直接在网上找。

## 1. 系统配置： ##

> OS: Ubuntu16.04  
> python Version: python3  
> CUDA Version: 10.0

## 2. caffe安装 ##

### 2.1 安装依赖库： ###

1.  `sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler`
2.  `sudo apt-get install --no-install-recommends libboost-all-dev`

### 2.2 下载caffe源码 ###

下载地址：[git clone https://github.com/BVLC/caffe.git][git clone https_github.com_BVLC_caffe.git]

## 3. 修改文件 ##

### 3.1 修改Makefile.config文件 ###

*  进入caffe源码主目录，拷贝一个Makefile.config作为编译配置选项

cp Makefile.config.example Makefile.config

*  修改编译文件Makefile.config

1.  `# USE_CUDNN := 1`
2.  `# USE_OPENCV := 0`
3.  `# OPENCV_VERSION := 3`
4.  `# WITH_PYTHON_LAYER := 1`
5.  
6.  `# 改为：`
7.  
8.  `USE_CUDNN := 1`
9.  `USE_OPENCV := 1`
10. `OPENCV_VERSION := 3`
11. `WITH_PYTHON_LAYER := 1`

1.  `INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include`
2.  `LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib`
3.  
4.  `改为：`
5.  
6.  `INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial`
7.  `LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial`

使用python3.5版本，把python2.7代码加注释，python3.5前的注释去掉

1.  `#PYTHON_INCLUDE := /usr/include/python2.7 \`
2.  `/usr/lib/python2.7/dist-packages/numpy/core/include`
3.  
4.  `PYTHON_INCLUDE := /usr/include/python3.5m \`
5.  `/usr/lib/python3.5/dist-packages/numpy/core/include`

### 3.2 修改Makefile文件 ###

1.  `NVCCFLAGS += -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)`
2.  `改为：`
3.  `NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)`
4.  
5.  
6.  `LIBRARIES += glog gflags protobuf boost_system boost_filesystem m`
7.  `改为：`
8.  `LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial`

## 4. 编译 ##

make all -j8

**遇到错误：**

**1. 编译选项问题**

> /usr/include/c++/4.9/bits/c++0x\_warning.h:32:2: error: \#error This file requires compiler and library support for the ISO C++ 2011 standard. This support is currently experimental, and must be enabled with the -std=c++11 or -std=gnu++11 compiler options.  
> \#error This file requires compiler and library support for the
> 
> ......
> 
> /usr/local/include/google/protobuf/arena.h:324:36: note: template argument deduction/substitution failed:  
> Makefile:600: recipe for target '.build\_release/src/caffe/proto/caffe.pb.o' failed  
> make: \*\*\* \[.build\_release/src/caffe/proto/caffe.pb.o\] Error 1

**解决方法：**请看这位大神是如何操作的[https://blog.csdn.net/pkwcxy/article/details/82351669?utm\_source=blogxgwz8][https_blog.csdn.net_pkwcxy_article_details_82351669_utm_source_blogxgwz8]

> 在Makefile文件中找到：
> 
> \#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#  
> \# Configure build  
> \#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#\#
> 
> 找到：
> 
> \# Linux
> 
> 添加：CXXFLAGS += -std=c++11

安装上面的方法一开始没报错，但过了一段时间后还是报同样的错误，再进一步如下修改，打开Makefile文件，给CXXFLAGS、NVCCFLAGS、LINKFLAGS都使用-std=c++11标准编译：

1.  `CXXFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS)`
2.  `改为：`
3.  `CXXFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11`
4.  
5.  
6.  `NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)`
7.  `改为：`
8.  `NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS) -std=c++11`
9.  
10. 
11. `LINKFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS)`
12. `改为：`
13. `LINKFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11`

**2. CUDA版本问题**

重新编译还是有错误：

> NVCC src/caffe/solvers/adagrad\_solver.cu  
> nvcc fatal : Unsupported gpu architecture 'compute\_20'  
> NVCC src/caffe/solvers/sgd\_solver.cu  
> nvcc fatal : Unsupported gpu architecture 'compute\_20'  
> Makefile:607: recipe for target '.build\_release/cuda/src/caffe/solvers/rmsprop\_solver.o' failed

原因是不支持GPU架构'compute\_20'，在Makefile.config里面的找到

CUDA\_ARCH := -gencode arch=compute\_20,code=sm\_20  
\-gencode arch=compute\_20,code=sm\_21 \\  
\-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\_60,code=sm\_60 \\  
\-gencode arch=compute\_61,code=sm\_61 \\  
\-gencode arch=compute\_61,code=compute\_61

把前面compute\_20相关的那2行删除  
\-gencode arch=compute\_20,code=sm\_20  
\-gencode arch=compute\_20,code=sm\_21

**3. g++版本问题**

> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> collect2: error: ld returned 1 exit status  
> Makefile:637: recipe for target '.build\_release/tools/upgrade\_solver\_proto\_text.bin' failed  
> make: \*\*\* \[.build\_release/tools/upgrade\_solver\_proto\_text.bin\] Error 1  
> make: \*\*\* 正在等待未完成的任务....  
> ......

这个g++版本问题，解决方法这里参考：[https://blog.csdn.net/chenshuibiao/article/details/78734957][https_blog.csdn.net_chenshuibiao_article_details_78734957]

先查看g++版本：

gcc --version     # 或 g++ --version

> gcc (Ubuntu 4.9.3-13ubuntu2) 4.9.3  
> Copyright (C) 2015 Free Software Foundation, Inc.  
> This is free software; see the source for copying conditions. There is NO  
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

发现我的是4.9.3版本，需要升级gcc版本。查看我Ubuntu系统发现里面已经存在更高级的5版本，就不用再安装了，直接强制软链接一下即可：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTI1MDU2MTc_size_16_color_FFFFFF_t_70][]

1.  `sudo ln -s /usr/bin/gcc-5 /usr/bin/gcc -f`
2.  `sudo ln -s /usr/bin/g++-5 /usr/bin/g++ -f`

再次查看g++版本发现已经变成5.4.0了

> g++ (Ubuntu 5.4.0-6ubuntu1~16.04.11) 5.4.0 20160609  
> Copyright (C) 2015 Free Software Foundation, Inc.  
> This is free software; see the source for copying conditions. There is NO  
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

再次编译caffe源码，发现没问题了。

**4. python版本问题**

如果在Makefile.config文件里面Python相关的配置有问题，可能会遇到以下的问题：

> CXX/LD -o .build\_release/tools/upgrade\_net\_proto\_text.bin  
> CXX/LD -o .build\_release/examples/siamese/convert\_mnist\_siamese\_data.bin  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/tools/convert\_imageset.o：在函数‘std::string\* google::MakeCheckOpString<unsigned long, int>(unsigned long const&, int const&, char const\*)’中：  
> convert\_imageset.cpp:(.text.\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/convert\_imageset.o：在函数‘main’中：  
> convert\_imageset.cpp:(.text.startup+0x347)：对‘google::SetUsageMessage(std::string const&)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/tools/caffe.o：在函数‘std::string\* google::MakeCheckOpString<cudaError, cudaError>(cudaError const&, cudaError const&, char const\*)’中：  
> caffe.cpp:(.text.\_ZN6google17MakeCheckOpStringI9cudaErrorS1\_EEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringI9cudaErrorS1\_EEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/caffe.o：在函数‘std::string\* google::MakeCheckOpString<unsigned long, int>(unsigned long const&, int const&, char const\*)’中：  
> caffe.cpp:(.text.\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/caffe.o：在函数‘main’中：  
> caffe.cpp:(.text.startup+0x53)：对‘google::SetVersionString(std::string const&)’未定义的引用  
> caffe.cpp:(.text.startup+0x86)：对‘google::SetUsageMessage(std::string const&)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/tools/extract\_features.o：在函数‘std::string\* google::MakeCheckOpString<int, int>(int const&, int const&, char const\*)’中：  
> extract\_features.cpp:(.text.\_ZN6google17MakeCheckOpStringIiiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringIiiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/extract\_features.o：在函数‘std::string\* google::MakeCheckOpString<unsigned long, unsigned long>(unsigned long const&, unsigned long const&, char const\*)’中：  
> extract\_features.cpp:(.text.\_ZN6google17MakeCheckOpStringImmEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringImmEEPSsRKT\_RKT0\_PKc\]+0x51)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/tools/compute\_image\_mean.o：在函数‘std::string\* google::MakeCheckOpString<int, int>(int const&, int const&, char const\*)’中：  
> compute\_image\_mean.cpp:(.text.\_ZN6google17MakeCheckOpStringIiiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringIiiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/compute\_image\_mean.o：在函数‘std::string\* google::MakeCheckOpString<unsigned long, int>(unsigned long const&, int const&, char const\*)’中：  
> compute\_image\_mean.cpp:(.text.\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringImiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/tools/compute\_image\_mean.o：在函数‘main’中：  
> compute\_image\_mean.cpp:(.text.startup+0x172)：对‘google::SetUsageMessage(std::string const&)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> .build\_release/examples/siamese/convert\_mnist\_siamese\_data.o：在函数‘convert\_dataset(char const\*, char const\*, char const\*)’中：  
> convert\_mnist\_siamese\_data.cpp:(.text+0x4cf)：对‘leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB\*\*)’未定义的引用  
> .build\_release/examples/siamese/convert\_mnist\_siamese\_data.o：在函数‘std::string\* google::MakeCheckOpString<unsigned int, int>(unsigned int const&, int const&, char const\*)’中：  
> convert\_mnist\_siamese\_data.cpp:(.text.\_ZN6google17MakeCheckOpStringIjiEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringIjiEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/examples/siamese/convert\_mnist\_siamese\_data.o：在函数‘std::string\* google::MakeCheckOpString<unsigned int, unsigned int>(unsigned int const&, unsigned int const&, char const\*)’中：  
> convert\_mnist\_siamese\_data.cpp:(.text.\_ZN6google17MakeCheckOpStringIjjEEPSsRKT\_RKT0\_PKc\[\_ZN6google17MakeCheckOpStringIjjEEPSsRKT\_RKT0\_PKc\]+0x50)：对‘google::base::CheckOpMessageBuilder::NewString()’未定义的引用  
> .build\_release/lib/libcaffe.so：对‘leveldb::Status::ToString() const’未定义的引用  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> Makefile:637: recipe for target '.build\_release/tools/compute\_image\_mean.bin' failed  
> collect2: error: ld returned 1 exit status  
> collect2: error: ld returned 1 exit status  
> make: \*\*\* \[.build\_release/tools/compute\_image\_mean.bin\] Error 1  
> make: \*\*\* 正在等待未完成的任务....  
> Makefile:637: recipe for target '.build\_release/tools/caffe.bin' failed  
> make: \*\*\* \[.build\_release/tools/caffe.bin\] Error 1  
> Makefile:637: recipe for target '.build\_release/tools/extract\_features.bin' failed  
> make: \*\*\* \[.build\_release/tools/extract\_features.bin\] Error 1  
> Makefile:637: recipe for target '.build\_release/tools/upgrade\_solver\_proto\_text.bin' failed  
> make: \*\*\* \[.build\_release/tools/upgrade\_solver\_proto\_text.bin\] Error 1  
> Makefile:637: recipe for target '.build\_release/tools/upgrade\_net\_proto\_binary.bin' failed  
> make: \*\*\* \[.build\_release/tools/upgrade\_net\_proto\_binary.bin\] Error 1  
> Makefile:637: recipe for target '.build\_release/tools/convert\_imageset.bin' failed  
> make: \*\*\* \[.build\_release/tools/convert\_imageset.bin\] Error 1  
> Makefile:637: recipe for target '.build\_release/tools/upgrade\_net\_proto\_text.bin' failed  
> make: \*\*\* \[.build\_release/tools/upgrade\_net\_proto\_text.bin\] Error 1  
> Makefile:642: recipe for target '.build\_release/examples/siamese/convert\_mnist\_siamese\_data.bin' failed  
> make: \*\*\* \[.build\_release/examples/siamese/convert\_mnist\_siamese\_data.bin\] Error 1  
>

经过自己研究发现，如果使用Python2版本就会有以上的问题，如果我改为 python3版本，即

PYTHON\_INCLUDE := /usr/include/python3.5m \\  
/usr/lib/python3.5/dist-packages/numpy/core/include

就发现没有问题了。我系统里面两个版本都安装有的，可能是默认使用python3的问题吧。

**5. 找不到 -lboost\_python3**

> CXX examples/cifar10/convert\_cifar\_data.cpp  
> AR -o .build\_release/lib/libcaffe.a  
> LD -o .build\_release/lib/libcaffe.so.1.0.0  
> /usr/bin/ld: 找不到 -lboost\_python3  
> collect2: error: ld returned 1 exit status  
> Makefile:585: recipe for target '.build\_release/lib/libcaffe.so.1.0.0' failed  
> make: \*\*\* \[.build\_release/lib/libcaffe.so.1.0.0\] Error 1

解决方法：

在/usr/lib/x86\_64-linux-gnu/ 下查看有没有libboost\_python-py3.5.so，如果有链接一下就可以了。

sudo ln -s libboost_python-py35.so libboost_python3.so

如果你使用的是python3.6版本，发现在 /usr/lib/x86\_64-linux-gnu/ 下没有libboost\_python-py3.6.so 的，那你就的自己编译一个了，可以参考我的另一篇博文：[https://blog.csdn.net/u012505617/article/details/88556621][https_blog.csdn.net_u012505617_article_details_88556621]

--------------------

参考：

[https://blog.csdn.net/sinat\_27240143/article/details/82701285][https_blog.csdn.net_sinat_27240143_article_details_82701285]

[https://blog.csdn.net/fanhenghui/article/details/80092131][https_blog.csdn.net_fanhenghui_article_details_80092131]

[git clone https_github.com_BVLC_caffe.git]: https://blog.csdn.net/u012505617/article/details/88532982?utm_medium=distribute.pc_relevant_ask_down.none-task-blog-blogcommendfrombaidu-5.nonecase&depth_1-utm_source=distribute.pc_relevant_ask_down.none-task-blog-blogcommendfrombaidu-5.nonecas
[https_blog.csdn.net_pkwcxy_article_details_82351669_utm_source_blogxgwz8]: https://blog.csdn.net/pkwcxy/article/details/82351669?utm_source=blogxgwz8
[https_blog.csdn.net_chenshuibiao_article_details_78734957]: https://blog.csdn.net/chenshuibiao/article/details/78734957
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTI1MDU2MTc_size_16_color_FFFFFF_t_70]: /images/20221124/cb579320aff64c258e8c3964c8d98714.png
[https_blog.csdn.net_u012505617_article_details_88556621]: https://blog.csdn.net/u012505617/article/details/88556621
[https_blog.csdn.net_sinat_27240143_article_details_82701285]: https://blog.csdn.net/sinat_27240143/article/details/82701285
[https_blog.csdn.net_fanhenghui_article_details_80092131]: https://blog.csdn.net/fanhenghui/article/details/80092131