mxnet环境搭建问题总结

为了跑maskrcnn搭建mxnet，gayhub地址如下：

https://github.com/TuSimple/mx-maskrcnn

源码编译安装参照如下官网：

https://mxnet.incubator.apache.org/install/build_from_source.html

https://mxnet.incubator.apache.org/install/centos_setup.html

 

下面开始记录各种问题的解决方法：

1.编译空间不足

fatal error: No space left on device

解决：

mkdir ~/tmp
export TMPDIR=~/tmp

参照如下网址：

https://stackoverflow.com/questions/31493663/unable-to-compile-with-make-fatal-error-no-space-left-on-device

 

2.找不到库

usr/bin/ld: cannot find -l

解决：

ld -lzlib --verbose

会有大概这样的输出：

==================================================
attempt to open /usr/x86_64-linux-gnu/lib64/libzlib.so failed
attempt to open /usr/x86_64-linux-gnu/lib64/libzlib.a failed
attempt to open /usr/local/lib64/libzlib.so failed
attempt to open /usr/local/lib64/libzlib.a failed
attempt to open /lib64/libzlib.so failed
attempt to open /lib64/libzlib.a failed
attempt to open /usr/lib64/libzlib.so failed
attempt to open /usr/lib64/libzlib.a failed
attempt to open /usr/x86_64-linux-gnu/lib/libzlib.so failed
attempt to open /usr/x86_64-linux-gnu/lib/libzlib.a failed
attempt to open /usr/local/lib/libzlib.so failed
attempt to open /usr/local/lib/libzlib.a failed
attempt to open /lib/libzlib.so failed
attempt to open /lib/libzlib.a failed
attempt to open /usr/lib/libzlib.so failed
attempt to open /usr/lib/libzlib.a failed
/usr/bin/ld.bfd.real: cannot find -lzlib

之后这样的命令解决：

sudo ln -s /usr/lib/libz.so.1.2.8 /usr/lib/libzlib.so

参照如下网址：

https://stackoverflow.com/questions/16710047/usr-bin-ld-cannot-find-lnameofthelibrary

 

3.找不到模块

undefined reference to imencode()

解决：

修改mxnet的Makefile文件

ifeq ($(USE_OPENCV), 1)
    LIBRARIES += opencv_core opencv_highgui opencv_imgproc 

    ifeq ($(OPENCV_VERSION), 3)
        LIBRARIES += opencv_imgcodecs
    endif

endif

参照如下两个网址：

https://blog.csdn.net/YhL_Leo/article/details/52150781

https://zhuanlan.zhihu.com/p/29820072

以及opencv的安装参考官网以及如下网址：

https://www.vultr.com/docs/how-to-install-opencv-on-centos-7

 

4.conda的虚拟环境下无法使用pip

pip freeze # now shows local packages!
pip install  # now installs a local package

即可。

参考如下网址：

https://github.com/ContinuumIO/anaconda-issues/issues/1429

 

5.CUDA相关

Traceback (most recent call last):
File "mxnet/example/image-classification/train_mnist.py", line 1, in 
import find_mxnet
File "/home/colin/mxnet/example/image-classification/find_mxnet.py", line 7, in 
import mxnet as mx
File "/home/colin/mxnet/example/image-classification/../../python/mxnet/init.py", line 7, in 
from .base import MXNetError
File "/home/colin/mxnet/example/image-classification/../../python/mxnet/base.py", line 43, in 
_LIB = _load_lib()
File "/home/colin/mxnet/example/image-classification/../../python/mxnet/base.py", line 34, in _load_lib
lib_path = libinfo.find_lib_path()
File "/home/colin/mxnet/example/image-classification/../../python/mxnet/libinfo.py", line 33, in find_lib_path
'List of candidates:\n' + str('\n'.join(dll_path)))
RuntimeError: Cannot find the files.
List of candidates:
/home/colin/mxnet/python/mxnet/libmxnet.so
/home/colin/mxnet/python/mxnet/../../lib/libmxnet.so

解决：

export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

参考如下网址：

https://github.com/apache/incubator-mxnet/issues/1174

 

6.opencv又找不到了

error while loading shared libraries: libopencv_core.so.2.4: cannot open shared object file: No such file or directory

解决：

You haven't put the shared library in a location where the loader can find it. look inside the /usr/local/opencv and /usr/local/opencv2 folders and see if either of them contains any shared libraries (files beginning in lib and usually ending in .so). when you find them, create a file called /etc/ld.so.conf.d/opencv.conf and write to it the paths to the folders where the libraries are stored, one per line.

for example, if the libraries were stored under /usr/local/opencv/libopencv_core.so.2.4 then I would write this to my opencv.conf file:

/usr/local/opencv/

Then run

sudo ldconfig -v

If you can't find the libraries, try running

sudo updatedb && locate libopencv_core.so.2.4

in a shell. You don't need to run updatedb if you've rebooted since compiling OpenCV.

参考如下网址：

https://stackoverflow.com/questions/12335848/opencv-program-compile-error-libopencv-core-so-2-4-cannot-open-shared-object-f