ubun_caffw_fast-rcnn编译坎坷路
记录一下今天这个伟大的时刻，搞了四天的fast终于环境打好了，模型跑起来了，还好我没放弃~唱起来，哈哈哈哈
先上几张结果显示，happy一下
下面将遇到的几个问题及解决方法拉出来，供大家分享：
编译caffe问题比较全的网站：
http://blog.sina.com.cn/s/blog_721a75e50102wfig.html
makefile.config no such file:
http://blog.****.net/jiajunlee/article/details/50373815#q4-makefileconfig-not-found
修改hdf5 fatal error：
http://blog.****.net/xue_wenyuan/article/details/52037121
Makefile:501: recipe for target '.build_debug/src/caffe/layers/tanh_layer.o' failed
make: *** [.build_debug/src/caffe/layers/tanh_layer.o] Error 1
先各种cudnn cp（http://blog.****.net/u011070171/article/details/52292680）不过，我们服务器里面不用cudnn，所以此步可以省略
又复制了caffe/include/layers文件夹到fast-rcnn
不管用。。。
按照组里已经配置好的MakeFile 和MakeFile.config配置的
哎呀，妈呀，终于装好了，是hdf5_hl下划线多了一个，什么找不到hdf5_hl（要看错误提示什么）
ImportError: numpy.core.multiarray failed to import
解决：
import numpy as np
np.path
['/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/numpy']（你自己的路径）
exit()
sudo rm -rf /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/numpy（上面的路径）
https://github.com/BVLC/caffe/issues/2092
先将./caffe-fast-rcnn/python/caffe/proto/caffe_pb2.py , all "systax = proto "，结果没作用。
。。
TypeError:__init__()got an unexpected keyword argument syntax\x03\x30.5\x12等错误
将/home/data/anaconda2文件夹拷贝我的目录下，更改gedit .bashrc文件,export path等，然后source，环境变量配置
（（（（注意：服务器上编程，应该第一步
第一步
第一步
设置环境变量，不然直接改动系统下东西（.bashrc，然后source更新）
最好将相应编程环境等先拷贝自己目录下））））
Traceback (most recent call last):
File "list_patients.py", line 13, in <module>
from skimage import measure,morphology
File "/home/nannan/scikit-image/skimage/measure/__init__.py", line 1, in <module>
from ._find_contours import find_contours
File "/home/nannan/scikit-image/skimage/measure/_find_contours.py", line 2, in <module>
from . import _find_contours_cy
ImportError: /home/nannan/scikit-image/skimage/measure/_find_contours_cy.so: undefined symbol: PyFPE_jbuf
解决：[email protected]:~$ python skimage_path.py
['/home/nannan/scikit-image/skimage']
[email protected]:~$ rm -rf /home/nannan/scikit-image/
make clean
bc: symbol lookup error: /home/data/anaconda2/lib/libreadline.so.6: undefined symbol: PC
已解决：
[email protected]:~/OBJ_detection/fast-rcnn/caffe-fast-rcnn$locate libreadline.so.6
/home/data/anaconda2/lib/libreadline.so.6
/home/data/anaconda2/lib/libreadline.so.6.2
/home/data/anaconda2/pkgs/readline-6.2-2/lib/libreadline.so.6
/home/data/anaconda2/pkgs/readline-6.2-2/lib/libreadline.so.6.2
/lib/x86_64-linux-gnu/libreadline.so.6
/lib/x86_64-linux-gnu/libreadline.so.6.3
[email protected]:~/OBJ_detection/fast-rcnn/caffe-fast-rcnn$ cp /lib/x86_64-linux-gnu/libreadline.so.6 ~/anaconda2/lib/
[email protected]:~/OBJ_detection/fast-rcnn/caffe-fast-rcnn$ make clean
bc: symbol lookup error: /home/data/anaconda2/lib/libreadline.so.6: undefined symbol: PC
发现没用。。。
[email protected]:~/OBJ_detection/fast-rcnn/caffe-fast-rcnn$ cp /lib/x86_64-linux-gnu/libreadline.so.6 /home/data/anaconda2/lib/
cp: cannot create regular file '/home/data/anaconda2/lib/libreadline.so.6': Permission denied
哎呀，发现不能轻易动服务器东西。。。
就想到吧.bashrc中环境变量，LD_LIBRAYRY变为我的/home/nannan/anaconda2.。。。
然后在cp系统中readline到我的环境下。好了，哈哈哈
./include/caffe/blob.hpp:9:34: fatal error: caffe/proto/caffe.pb.h: No such file or directory
compilation terminated.
Makefile:501: recipe for target '.build_release/src/caffe/layers/threshold_layer.o' failed
解决：
[email protected]:~/caffe/src/caffe/proto$ protoc --cpp_out=/home/nannan/OBJ_detection/fast-rcnn/caffe-fast-rcnn/include/caffe/ caffe.proto
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
/usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to `[email protected]_5.0'
解决：
这是由于ａｎａｃｏｎｄａ的ｌｉｂlzma.so 与 /lib/x86_64-linux-gnu/中的ｌｉｂｌｚｍａ.so发生冲突，需要在编译时将ａｎａｃｏｎｄａ的ｌｉｂ路径注释掉# export #LD_LIBRARY_PATH="/home/dl/anaconda2/lib:$LD_LIBRARY_PATH"
加上：
export LD_LIBRARY_PATH=/lib/x86_64-linux-gnu/
在/home/bigtop/py-faster-rcnn/tools目录下运行, python demo.py --cpu此时运行到最后可能会出现内存溢出现象，添加执行参数 --net caffenet 即采用小型模型进行训练。