1)Fatal error : 'tr1/tuple' file not found
出现该问题有两种情况,可以先尝试下面的链接:https://github.com/BVLC/caffe/issues/1358 如果不行,那说明是 Makefile
文件除了问题。一般来说,按照 https://github.com/BVLC/caffe/pull/1740里的 33a56e0
那个 post 来修改 Makefile
文件就解决了。 在出错之前,Makefile
相关内容如下:
ifeq ($(OSX), 1)
CXX := /usr/bin/clang++
CXXFLAGS += -stdlib=libstdc++
LINKFLAGS += -stdlib=libstdc++
# clang throws this warning for cuda headers
WARNINGS += -Wno-unneeded-internal-declaration
ifneq ($(findstring 10.10, $(shell sw_vers -productVersion)),)
CXXFLAGS += -stdlib=libc++
LINKFLAGS += -stdlib=libc++
endif
# gtest needs to use its own tuple to not conflict with clang
CXXFLAGS += -DGTEST_USE_OWN_TR1_TUPLE=1
# boost::thread is called boost_thread-mt to mark multithreading on OS X
LIBRARIES += boost_thread-mt
NVCCFLAGS += -DOSX
然后修改成 33a56e0
的样子就成功了。
2) make: *** [matlab/caffe/caffe.mexmaci64] Error 255
解决方法 https://github.com/BVLC/caffe/issues/1212
3) make: *** [.build_release/tools/caffe.bin] Error 1
解决方法 存在上一次安装的残留文件。用 make clean
清除之前的安装,重新编译即可
4) make: *** [runtest] Trace/BPT trap: 5
解决方法:adna 没有安装 hdf5
。重新安装 hdf5
。https://github.com/BVLC/caffe/issues/454
5) 错误:
Building with 'Xcode Clang++'.
Undefined symbols for architecture x86_64:
"std::string::find(char, unsigned long) const", referenced from:
boost::basic_format<char, std::char_traits<char>, std::allocator<char> >::parse(std::string const&) in libcaffe.a(math_functions.o)
int boost::io::detail::upper_bound_from_fstring<std::string, std::ctype<char> >(std::string const&, std::string::value_type, std::ctype<char> const&, unsigned char) in libcaffe.a(math_functions.o)
"std::string::compare(char const*) const", referenced from:
_mexFunction in matcaffe.o
caffe::UpgradeV0PaddingLayers(caffe::NetParameter const&, caffe::NetParameter*) in libcaffe.a(upgrade_proto.o)
caffe::UpgradeLayerParameter(caffe::LayerParameter const&, caffe::LayerParameter*) in libcaffe.a(upgrade_proto.o)
caffe::UpgradeV0LayerType(std::string const&) in libcaffe.a(upgrade_proto.o)
caffe::Filler<float>* caffe::GetFiller<float>(caffe::FillerParameter const&) in libcaffe.a(dummy_data_layer.o)
caffe::Filler<double>* caffe::GetFiller<double>(caffe::FillerParameter const&) in libcaffe.a(dummy_data_layer.o)
caffe::WindowDataLayer<float>::DataLayerSetUp(std::vector<caffe::Blob<float>*, std::allocator<caffe::Blob<float>*> > const&, std::vector<caffe::Blob<float>*, std::allocator<caffe::Blob<float>*> >*) in libcaffe.a(window_data_layer.o)
...
解决方法: https://github.com/BVLC/caffe/issues/1212 再看->https://github.com/BVLC/caffe/pull/1310 然后看->https://github.com/CellScope/caffe/commit/954ac8f1fca889ada7194188db2cda3c300b0be0
在 makefile
中做如下更改
$(MATLAB_DIR)/bin/mex $(MAT$(PROJECT)_SRC) \
CXX="$(CXX)" \
CXXFLAGS="\$$CXXFLAGS $(MATLAB_CXXFLAGS)" \
- CXXLIBS="\$$CXXLIBS $(STATIC_LINK_COMMAND) $(LDFLAGS)" -output $@
+ CXXLIBS="\$$CXXLIBS $(STATIC_LINK_COMMAND) $(LDFLAGS) /usr/lib/libstdc++.dylib" -output $@
6)错误:
Building with 'Xcode Clang++'.
Undefined symbols for architecture x86_64:
"_mxArrayToString", referenced from:
init(int, mxArray_tag**, int, mxArray_tag const**) in matcaffe.o
read_mean(int, mxArray_tag**, int, mxArray_tag const**) in matcaffe.o
_mexFunction in matcaffe.o
"_mxCreateCellArray_700", referenced from:
get_weights(int, mxArray_tag**, int, mxArray_tag const**) in matcaffe.o
解决方法: https://github.com/BVLC/caffe/issues/915 做两个改动, 一、
change the LIBRARY_DIRS
section of Makefile.config
to read:LIBRARY_DIRS := $(PYTHON_LIB) /Applications/MATLAB_R2014a.app/bin/maci64 /usr/local/lib /usr/lib
二、
change the mexopts.sh
to include 10.10
wherever 10.7
was there ( 4 places).
7) 按下述指引进行设置 makefile
https://github.com/BVLC/caffe/pull/1740 ->https://github.com/shelhamer/caffe/commit/ab839f5b2f5c93da34c2ab797ab2a64b62645976
8)错误:
Building with 'Xcode Clang++'.
Undefined symbols for architecture x86_64:
"google::protobuf::io::CodedOutputStream::WriteStringWithSizeToArray(std::string const&, unsigned char*)", referenced from:
caffe::Datum::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::FillerParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::SolverParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::SolverState::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetState::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetStateRule::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
解决办法:出处忘记了... @reking, I see boost errors, which I vaguely recall getting myself. If I recall, Caffe was seeing Matlab's internal libraries earlier in the library path than my local homebrew libraries in /usr/local/lib. Caffe might be trying to link Matlab's version of the boost library, which, needless to say, isn't compatible with Caffe.
Try changing the LIBRARY_DIRS := ... line in your Makefile.config so that the /usr/local/lib
directory is before your Matlab library directory.
Mine looks like this:
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib $(MATLAB_DIR)/bin/maci64 /usr/lib
9) 错误:
Building with 'Xcode Clang++'.
Undefined symbols for architecture x86_64:
"google::protobuf::io::CodedOutputStream::WriteStringWithSizeToArray(std::string const&, unsigned char*)", referenced from:
caffe::Datum::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::FillerParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::SolverParameter::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::SolverState::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetState::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
caffe::NetStateRule::SerializeWithCachedSizesToArray(unsigned char*) const in libcaffe.a(caffe.pb.o)
解决办法: 9) 的错误和 8)很相似。但是我注意到错误提示: clang++ .build_release/src/caffe/proto/caffe.pb.cc -stdlib=libstdc++ -DGTEST_USE_OWN_TR1_TUPLE=1
有这么一行。在这一行里,有 -stdlib=libstdc++
字段。如果其他依赖库用 libc++ 编译的话,那么就会出现 9)的错误。解决办法是在 Makefile 文件里,将带有 libstdc++ 的行注释掉即可。
2015.7.2 补充。今天在编译老版本 caffe-rc2
版本的时候,又出现了一次该错误。但当时我是把最新版本的 makefile.config
文件给复制过来用了,并没有重新填写。所以,考虑到这个情况。我又重新解压了一次 caffe-rc2.zip 压缩包,重新手动填写了一次 makefile.config
文件。再次编译就好了。
10) Error with mexopencv
出现这种错误的原因解释在下面链接的 OSX 一节:http://kyamagu.github.io/mexopencv/ 解决方法上面链接也给了。但是下面的链接讨论的更为详细:https://github.com/kyamagu/mexopencv/issues/71
我自己的解决办法是,在 terminal
里输入DYLD_INSERT_LIBRARIES=/usr/local/lib/libopencv_highgui.2.4.dylib:/usr/local/lib/libtiff.5.dylib /Applications/MATLAB_R2014b.app/bin/matlab
因为,matlab
运行 RCNN
代码时提示 libopencv_highgui.2.4.dylib
使用的是 Matlab
自己的。所以就在终端里插入系统库。 而libopencv_highgui.2.4.dylib
系统库又要调用 libtiff.5.dylib
, 所以再继续插入libtiff.5.dylib
。
11) Issue with libopenblas.so.0
在编译 caffe
中的最后一步,运行 make runtest
时,出现了 caffe
无法找到libopenblas.so.0
的错误。出现这个错误的原因是,从源码编译 openblas
时手动更改了安装路径。 解决方法是,建立一个软连接到 openblas
的默认安装路径即可。 cd /opt sudo ln -s /usr/local/OpenBLAS/ .
然后将 openblas
的库所在位置添加到系统环境变量 LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/opt/OpenBLAS/lib/ sudo ldconfig
这时,在编译就不会出错了。
参考 https://github.com/sermanet/OverFeat/issues/10
12) 找不到 GLIBCXX_3.4.20 文件
Invalid MEX-file '/home/coldmoon/ComputerVision/Caffe/matlab/caffe/caffe.mexa64':
/home/coldmoon/MATLAB/R2014b/bin/glnxa64/../../sys/os/glnxa64/libstdc++.so.6: version `GLIBCXX_3.4.20'
not found (required by /home/coldmoon/ComputerVision/Caffe/matlab/caffe/caffe.mexa64)
参考 http://askubuntu.com/questions/575505/glibcxx-3-4-20-not-found-how-to-fix-this-error http://*.com/questions/16605623/where-can-i-get-a-copy-of-the-file-libstdc-so-6-0-15
上面两个链接指出的是,在 libstdc++.so.6
中无法找到 GLIBCXX_3.4.20
时的解决办法。但在我系统里 OSX(10.10),情况跟上述不一样。 通过下列命令 find / -name "libstdc++.so.6"
可以找到官方提供的库所在路径。 然后进入该路径: strings ./libstdc++.so.6 | grep GLIBCXX
可以看到,官方提供的 libstdc++.so.6
已经包含了 GLIBCXX_3.4.20
。这说明 caffe.mexa64
所使用的库并非官方的库,而是 matlab
自己提供的库 根据https://github.com/rbgirshick/rcnn/issues/13 的说法,这是LD_LIBRARY_PATH
设置不当造成的。导致了程序优先寻找 matlab
目录下的 libstdc++.so.6
。至于 matlab
究竟引用了哪里的库,可以通过 ldd
命令查看。 根据 https://github.com/rbgirshick/rcnn/issues/9 在终端下,输入ldd caffe.mexa64
,可以看到一堆所引用的库路径。而在 matlab
的命令窗口中输入 !ldd caffe.mexa64
则可以看到 matlab
运行 caffe
函数时,究竟在引用哪些库。不出意外的发现,matlab
引用的库路径果真都是 matlab
自己的。因为其中列出一条: libstdc++.so.6 => /home/coldmoon/MATLAB/R2014b/sys/os/glnxa64/libstdc++.so.6 (0x00007f2664bb6000)
再次运行 strings ./libstdc++.so.6 | grep GLIBCXX
,就可以发现,果真没有 GLIBCXX_3.4.20
.
解决方法,参考 https://github.com/BVLC/caffe/issues/655https://github.com/kyamagu/mexopencv/issues/64https://github.com/kyamagu/mexopencv/issues/62#issuecomment-15054244
当我把所有 !ldd caffe.mexa64
输出的结果都放到 LD_PRELOAD
中是,!ldd
这个命令会出错,即使不给它参数也会出错。因此, 我只把libstdc++
放到了这个环境变量里。最终形成一个脚本文件来运行 matlab
:
#!/bin/bash
export LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libstdc++.so.6
echo "LDPRELOAD is:"
echo $LD_PRELOAD
/home/coldmoon/MATLAB/R2014b/bin/matlab