[caffe安装]配置环境过程中出现的问题及解决

 今天要跑一下Convolutional Autoencoder for Loop Closure，轻量级神经网络闭环方法。

caffe安好之后编译程序出现以下错误：

Scanning dependencies of target deeplcd
[  5%] Building CXX object CMakeFiles/deeplcd.dir/src/deeplcd/deeplcd.cpp.o
In file included from /home/lenovo/litw/dependencies/caffe/build/install/include/caffe/common.hpp:19:0,
                 from /home/lenovo/litw/dependencies/caffe/build/install/include/caffe/blob.hpp:8,
                 from /home/lenovo/litw/dependencies/caffe/build/install/include/caffe/caffe.hpp:7,
                 from /home/lenovo/litw/DL/calc/DeepLCD/include/deeplcd/deeplcd.h:1,
                 from /home/lenovo/litw/DL/calc/DeepLCD/src/deeplcd/deeplcd.cpp:1:
/home/lenovo/litw/dependencies/caffe/build/install/include/caffe/util/device_alternate.hpp:34:10: fatal error: cublas_v2.h: 没有那个文件或目录
 #include <cublas_v2.h>
          ^~~~~~~~~~~~~
compilation terminated.
CMakeFiles/deeplcd.dir/build.make:62: recipe for target 'CMakeFiles/deeplcd.dir/src/deeplcd/deeplcd.cpp.o' failed
make[2]: *** [CMakeFiles/deeplcd.dir/src/deeplcd/deeplcd.cpp.o] Error 1
CMakeFiles/Makefile2:136: recipe for target 'CMakeFiles/deeplcd.dir/all' failed
make[1]: *** [CMakeFiles/deeplcd.dir/all] Error 2
Makefile:94: recipe for target 'all' failed
make: *** [all] Error 2

大概是因为缺少cuda，但安装cuda失败，提示缺少kernel module，阅读CUDA安装说明才发现（怎么这个时候才读啦），我电脑上根本就没有nvidia GPU...回过头来看CUDA只是程序的可选依赖项，最后发现是因为在安装caffe时没有把设置改成cpu-only.

然后在caffe包里的CMakeLists.txt里面CPU-ONLY这一参数设置为ON才可。但还是程序编译失败了。

然后使用推荐的make编译。先需要修改Makefile.config文件，去注释掉cpu-only那一行。然后出现错误：

In file included from src/caffe/solvers/sgd_solver.cpp:5:0:
./include/caffe/util/hdf5.hpp:7:10: fatal error: hdf5.h: 没有那个文件或目录
 #include "hdf5.h"
          ^~~~~~~~
compilation terminated.
Makefile:591: recipe for target '.build_release/src/caffe/solvers/sgd_solver.o' failed
make: *** [.build_release/src/caffe/solvers/sgd_solver.o] Error 1

按照https://blog.csdn.net/object_allen/article/details/73933934里面的方法，改进了一下，

对原文行不做修改，而是对197,198行做修改

ifeq ($(USE_HDF5), 1)
	LIBRARIES += hdf5_hl hdf5
把上面两行改成
ifeq ($(USE_HDF5), 1)
	LIBRARIES += hdf5_serial_hl hdf5_serial

大概是ubuntu16.04之后更新了路径

然后又出错辽，因为caffe默认opencv版本是2.x，但我装的3.4.4

报错如下：

.build_release/lib/libcaffe.so：对‘cv::imread(cv::String const&, int)’未定义的引用
.build_release/lib/libcaffe.so：对‘cv::imencode(cv::String const&, cv::_InputArray const&, std::vector<unsigned char, std::allocator<unsigned char> >&, std::vector<int, std::allocator<int> > const&)’未定义的引用
.build_release/lib/libcaffe.so：对‘cv::imdecode(cv::_InputArray const&, int)’未定义的引用
collect2: error: ld returned 1 exit status
Makefile:635: recipe for target '.build_release/tools/upgrade_solver_proto_text.bin' failed
make: *** [.build_release/tools/upgrade_solver_proto_text.bin] Error 1

在https://blog.csdn.net/u011636440/article/details/82660697找到解决方案，把Makefile.config中OPENCV_VERSION := 3取消注释即可。

但最后还是删掉重新编译成功，

but you know what,

还是开头这个错误！但既然caffe是编译成功的，所以是程序设置的问题。在https://blog.csdn.net/xcls2010/article/details/80422538中找到解决方案，给CXX_FLAGS += -DCPU_ONLY=1 这一项即可。

然后继续跑又出错了（崩溃边缘）（幸好之前就有看到解决方案）

[ 16%] Building CXX object CMakeFiles/deeplcd.dir/src/deeplcd/deeplcd.cpp.o
In file included from /home/lenovo/litw/dependencies/caffe/include/caffe/caffe.hpp:7:0,
                 from /home/lenovo/litw/DL/calc/DeepLCD/include/deeplcd/deeplcd.h:1,
                 from /home/lenovo/litw/DL/calc/DeepLCD/src/deeplcd/deeplcd.cpp:1:
/home/lenovo/litw/dependencies/caffe/include/caffe/blob.hpp:9:10: fatal error: caffe/proto/caffe.pb.h: 没有那个文件或目录
 #include "caffe/proto/caffe.pb.h"
          ^~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
CMakeFiles/deeplcd.dir/build.make:62: recipe for target 'CMakeFiles/deeplcd.dir/src/deeplcd/deeplcd.cpp.o' failed