Caffe installation is very troublesome, especially my last choice is to install on Windows, really not easy.
Caffe in Ubuntu installation is relatively simple, a lot of information, later installed and then write the installation process.
Here are some references to the Caffe installation process on Windows.
Https://github.com/happynear/caffe-windows
Https://github.com/niuzhiheng/caffe
http://blog.csdn.net/fengbingchun/article/details/47905907
Http://www.bubuko.com/infodetail-902302.html
https://initialneil.wordpress.com/2015/01/11/ build-caffe-in-windows-with-visual-studio-2013-cuda-6-5-opencv-2-4-9/#comment-48
Http://www.aiuxian.com/article/p-2410195.html
If you're already feeling big, it's okay, dear. Microsoft helped us sort out a caffe version, which is simply a one-click installation.
Project Address Https://github.com/Microsoft/caffe?url_type=39&object_type=webpage&pos=1
My installation Environment VS2013 + Win7 64-bit
Briefly describe the installation process.
1. Suppose you extracted the directory called Caffe-microsoft
Then copy the Caffe-microsoft\windows\commonsettings.props.example to a part called Caffe-microsoft\windows\. Commonsettings.props, which is the configuration file we need.
Open for editing, it is recommended to edit with notepad++.
The configuration is very simple, the basic look can guess what is done.
Choose whether to use Cpu_only mode or Cuda. The CPU only uses the CPU only, if using the GPU, it will need to install CUDA driver, configure the relevant version.
Later choose whether to compile Python and matlab.
If CUDNN is used, the associated path is configured.
The same is true for Python and Matlab.
Because my Python installation is 32 bits and cannot be compiled, support for MATLAB and Python is not turned on. You can refer to Https://github.com/Microsoft/caffe?url_type=39&object_type=webpage&pos=1 for your own reference and write it in quite detail.
By the way, if you turn on Python support, but the compilation error is as follows
Then it should be that your Python is not 64-bit. Consider reloading.
2. All you need to do is open the caffe-microsoft\windows\ Caffe.sln, into the VS2013 project, and then compiled, he will download a large number of dependent libraries, in Caffe-microsoft's parent directory, will generate a nugetpackages directory, loaded is a variety of dependent libraries. When everything is done, the build will be fine.
For the use of C + + interfaces, you can refer to Caffe-microsoft\examples\cpp_classification\classification.cpp
The whole process is simple ~
PS. Interested can look at the various configuration files in this project, the personal feel is simply art, can study well.