Win10下搭建Mxnet并训练handwritten digits_handwritten uci-优快云博客

本文链接：https://blog.youkuaiyun.com/enight0000/article/details/52172852

本文详细介绍了在Windows环境下如何编译Mxnet，并提供了一份手写数字识别的实战案例。包括配置Visual Studio 2013以支持更多C++11特性、安装第三方库如OpenCV和CuDNN等步骤。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Mxnet下的中文教程不多，希望大家能多多交流

http://note.youdao.com/yws/public/redirect/share?id=9a0f37906a34912095943f4932a5bbaa&type=false

允许转载但必须保留原文链接信息，如若贴图、链接丢失，请点击上方阅读原文

Leon Xie

最新修改日期 2016/08/10

Windows下提供了预编译包，会方便很多，使用预编译包的请跳过前两步

预备程序

Firstly, we should make your Visual Studio 2013 support more C++11 features.

- Download and install Visual C++ Compiler Nov 2013 CTP.
- Copy all files in C:\Program Files (x86)\Microsoft Visual C++ Compiler Nov 2013 CTP (or the folder where you extracted the zip archive) to C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC and overwrite all existed files. Don’t forget to backup the original files before copying.

Secondly, fetch the third-party libraries, including OpenCV, CuDNN and OpenBlas(ignore this if you have MKL).

- NOTICE: You need to register as a NVIDIA community user to get the download link of CuDNN.

Finally, use CMake to create a Visual Studio solution in ./build/. During configuration, you may need to set the path of each third-party library, until no error is reported. (Set environmental variable OpenBLAS_HOME to the OpenBLAS directory containing include and lib; set OpenCV_DIR to thebuild directory after unpacking the OpenCV package.) Open the solution and compile, you will get amxnet.dll in ./build/Release or ./build/Debug.

重点在这里：

1.如果你已经装了VS2015可以不用安装Visual C++ Compiler Nov 2013 CTP.

~~2.CuDNN需要下载V3版本(0.7版支持CuDnn v5)，也就是cudnn-7.0-win-x64-v3.0-prod~~

3.CUDA需要下载最新8.0，7.5虽然兼容Win10但无法完成Mxnet的编译。

4.你需要安装CMake这个软件 cmake-3.6.1.zip

5.你需要把Mxnet项目从github上clone下来，但是国内clone好慢而且老挂，所以我们必然用Download Zip，但由于它引用了其他项目，下载后会发现引用的项目空有一个文件夹，那就把其他项目也打包下载下来，然后手动解压咯。

编译过程

1.把Mxnet项目解压完毕(引用项目也解压好)

2.建文件夹build

3.建一个build.bat 里面写

cmake -G "Visual Studio 12 2013 Win64" ..

cmake-gui .

如果你用VS2015，改第一个参数为Visual Studio 14 2015 Win64

4.你需要把cmake/bin添加到系统PATH中

5.运行build.bat，修改各种参数，点Configure直到通过，放一张图（图上CUDA的路径不对，要用8.0的）

6.you will get amxnet.dll in ./build/Release or ./build/Debug 这就是我们要的东西

7.Customized Building

The configuration of mxnet can be modified by config.mk

modify the compiling options such as compilers, CUDA, CUDNN, Intel MKL, various distributed filesystem such as HDFS/Amazon S3/...
First copy make/config.mk to the project root, on which any local modification will be ignored by git, then modify the according flags.

Building with Intel MKL Support

First, source /path/to/intel/bin/compilervars.sh to automatically set environment variables. Then, edit make/config.mk, let USE_BLAS = mkl. USE_INTEL_PATH = NONE is usually not necessary to be modified.

训练handwritten digits

这里以pre-built packages on Window项目进行说明。

1.安装python，2.7以上均可，推荐Windows x86-64 executable installer

2.安装python库numpy，虽然Mxnet提供了py脚本来自动安装，但是由于VS版本问题，各种编译失败，所以直接下载预编译numpy进行安装。

You can find there several wheels for numpy: http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy. To get the proper architecture, check in the name win32 for 32 bits and amd64 for 64 bits. To get the proper Python version, check cpXX: first X is major version, and second X is minor version, so for instance cp27 means CPython 2.7.

Example: pip install numpy‑1.9.2rc1+mkl‑cp27‑none‑win32.whl

对于输入pip提示命令不存在什么的童鞋，请将Python27\Scripts加入系统PATH