Compile Caffe with Visual Studio 2013 on Windows 7 x64, using Cuda 7.0, boost 1.57.0, OpenCV 2.4.11 and OpenBlas 0.2.14.

After succesfully build, you will get tools like caffe.exe and examples like convert_mnist_data.exe under path bin, which can be used for Train-and-Test on the dataset mnist.

Prerequisites

Pre-Build

### Install Visual Studio, CUDA, boost, OpenCV, and OpenBlas.
Set up $(CUDA_PATH), $(OPENCV_PATH), $(BOOST_PATH)

Compile GFlags, GLog, ProtoBuf, LevelDB, HDF5, and LMDB from source.

Then put then under the header and library filesmsvc/3rdparty folder like this.

	3rdparty
	├─gflags
	│  ├─include
	│  │  └─gflags
	│  └─x64
	│      └─Release
	├─glog
	│  ├─glog
	│  └─x64
	│      └─Release
	├─HDF5
	│  ├─bin
	│  ├─include
	│  └─lib
	├─leveldb
	│  ├─include
	│  │  └─leveldb
	│  └─x64
	│      └─Release
	├─lmdb
	│  ├─include
	│  └─x64
	│      └─Release
	├─openblas
	│  ├─bin
	│  ├─include
	│  └─lib
	├─port
	└─protobuf
	    ├─include
	    │  └─google
	    │      └─protobuf
	    │          ├─compiler
	    │          │  ├─cpp
	    │          │  ├─java
	    │          │  ├─javanano
	    │          │  ├─python
	    │          │  └─ruby
	    │          ├─io
	    │          └─stubs
	    └─x64
	        └─Release

Generate proto file

Create a bash file named GeneratePB.bat under the 3rdparty folder, write as follow 

cd /d %~dp0

echo off

set PROTO_PATH=../../src/caffe/proto
set PROTOC="protobuf\x64\Release\protoc"

if exist "%PROTO_PATH%/caffe.pb.h" (
    echo caffe.pb.h remains the same as before
) else (
    echo caffe.pb.h is being generated
    %PROTOC% -I="%PROTO_PATH%" --cpp_out="%PROTO_PATH%" "%PROTO_PATH%/caffe.proto"
)

if exist "%PROTO_PATH%/caffe_pretty_print.pb.h" (
    echo caffe_pretty_print.pb.h remains the same as before
) else (
    echo caffe_pretty_print.pb.h is being generated
    %PROTOC% -I="%PROTO_PATH%" --cpp_out="%PROTO_PATH%" "%PROTO_PATH%/caffe_pretty_print.proto"
)

echo "Done!"
pause

Double click it, to get caffe.*.* files under src/caffe/proto/ .

Additional Files

There are some additional files under 3rdparty\port\ for porting to windows version. You may include then to the project.
getopt.c getopt.h mkstemp.cpp mkstemp.h unistd.h

Build

  • Open caffe.sln.
  • Change the platform from x64 to Build -> Configuration Manager -> Active solution platform -> new -> x64(Win32) -> OK
  • Change lib path to Additional Library Directories _in _Configuration Properties -> Linker -> General: (Both Debug and Release)
  • Change include path to Additional Library Directories _in _Configuration Properties -> C/C++ -> General: (Both Debug and Release)
  • Modify compute_30,sm_30 in Configuration Properties -> Cuda C++ -> Device -> Code Generation upto your device.
  • Add lib path to Additional Library Directories _in _Configuration Properties -> Linker -> General: (Both Debug and Release) [Change the right version upto your opencv]
opencv_core2411.lib
opencv_flann2411.lib
opencv_imgproc2411.lib
opencv_highgui2411.lib
opencv_legacy2411.lib
opencv_video2411.lib
opencv_ml2411.lib
opencv_calib3d2411.lib
opencv_objdetect2411.lib
opencv_stitching2411.lib
opencv_gpu2411.lib
opencv_nonfree2411.lib
opencv_features2d2411.lib
cudart.lib
cuda.lib
nppi.lib
cufft.lib
cublas.lib
curand.lib
gflags.lib
libglog_static.lib
libopenblas.dll.a
libprotobuf.lib
libprotoc.lib
leveldb.lib
lmdb.lib
libhdf5.lib
libhdf5_hl.lib
Shlwapi.lib
  • Add Flags to _Configuration Properties -> C/C++ -> Command Line _(Both Debug and Release mode) 
_CRT_SECURE_NO_WARNINGS
-D_SCL_SECURE_NO_WARNINGS
GOOGLE_GLOG_DLL_DECL=

Ok, start build.

Main Errors may encoutered.

open		_open
close		_close
snprintf	sprintf_s
getpid		_getpid

To be continued.

Thanks

License and Citation

Caffe is released under the BSD 2-Clause license.
The BVLC reference models are released for unrestricted use.

Please cite Caffe in your publications if it helps your research:

@article{jia2014caffe,
  Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
  Journal = {arXiv preprint arXiv:1408.5093},
  Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
  Year = {2014}
}

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution 4.0 International License.
A permanent link is required as http://mthust.github.io/2015/04/Build-Caffe-with-Visual-Studio-2013-on-Windows-7-x64-using-Cuda-7.0