MTCNN简单实践（Windows下happynear代码）

我的相关博客：

《Windows 7+Visual Studio 2015下Caffe安装（CPU/非GPU）》

《Windows 7+Visual Studio 2015下Caffe安装（GPU+Cuda 9.0+cuDNN 7.3）》

---

在初步安装好Caffe-CPU后，准备尝试实践一下MTCNN。

本机环境：笔记本、GTX970M的GPU、windows 7、VS2015、python 2.7、MATLAB R2015b。

参考博客从happynear的github上下载测试程序进行测试。

打开code/codes/vs/下的vs工程，一共有四个项目：MatMTCNN（生成可在Matlab中调用的库），PyCascadeFaceDetection（生成可在python中调用的库），TestCascadeCNN（生成exe可执行文件），testPyCascadeCNN（py文件，测试PyCascadeFaceDetection，vs中无法加载，可以删掉或不管）。

先试一下生成（Debug模式）前3个，都失败了。

一个一个来，matlab调用、python调用先不管，主要关注TestCascadeCNN，参考博客里的，将各个库、包含文件的路径进行配置。方便一下大家Copy。

TestCascadeCNN添加包含目录和库目录如下：
Include Directories
E:\study\ML\caffe-windows-ms\include
E:\study\ML\caffe-windows-ms\src
E:\study\ML\caffe-windows-ms\src\caffe
E:\study\ML\caffe-windows-ms\windows\caffe.binding
E:\study\ML\caffe-windows-ms\windows\thirdparty\Boost
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenCV\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenCV\include\opencv
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenCV\include\opencv2
E:\study\ML\caffe-windows-ms\windows\thirdparty\GFlags\Include
E:\study\ML\caffe-windows-ms\windows\thirdparty\GLog\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\HDF5\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\LEVELDB\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\LMDB\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenBLAS\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\Protobuf\include

Library Directories
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenCV\x64\vc14\lib

Additional Include Directories
E:\study\ML\caffe-windows-ms\windows\caffe.binding
E:\study\ML\caffe-windows-ms\windows\thirdparty\HDF5\include
E:\study\ML\caffe-windows-ms\windows\thirdparty\Boost
E:\study\ML\caffe-windows-ms\Build\x64\Release
E:\study\ML\caffe-windows-ms\windows\thirdparty\OpenCV\x64\vc14\bin

Additional Library Directories
E:\study\ML\caffe-windows-ms\windows\thirdparty\HDF5\lib 
E:\study\ML\caffe-windows-ms\windows\thirdparty\Ope