Skip to content
cpb8010 edited this page Jul 9, 2013 · 1 revision

These directions are assuming the project is being setup on Windows 7 or 8. Linux and OSX should work, but have not yet been tested.

Recommended Prerequisite tools

It's just C++ and Python, but the project is already setup to use the below

  • Visual Studio Express 2012 (C++). 32 or 64 bit.

Required libraries

These are required to compile and run the project

  • Boost v1.48 or above

Make or set an environment variable named BOOST_ROOT to the folder containing the complied Boost library. This is expected by visual studio. E.g. setx BOOT_ROOT "C:\boost\boost_1_54"

  • OpenCV 2.4.0 or above

Make or set an environment variable called OPENCV_PATH to the folder containing the compiled OpenCV package. This is expected by the VS project settings.

  • YAML-CPP 0.5.1

This is only required by the example programs to specify advanced fitting options, there are currently no switches to remove this requirement, but it could be done manually. The Visual Studio project expects YAMLCPP_PATH as an environment variable.

  • Python 2.7.5 or above

Using older versions will cause problems during result parsing because of a memory leak with os.IsDir Python 2.7 was chosen because other required libraries had better support for it on Windows. Clearly, install this first before trying to install other Python packages. Don't forget to add to PATH to make things easier.

  • NumPy and SciPy
    • SciKit-Learn

      Only used to generate random cross-validation fold pairing, under-utilized but provides many more tools to expand later.

    • Matplotlib

      • Pandas

Common Issues

  1. Linking/Include errors with OpenCV and/or Boost

Compare the environment variable with how the path is being specified in the include or library options. If all else fails, try hard-coding your include path in the visual studio options and dll directories to uncover other possible issues with the configuration.

  1. Python can't find cv module

Don't forget to add this to your Python path, the OpenCV installer might not find it for you.

  1. There are so many warnings I can't see what the compile errors are!

These are mostly warnings about using C functions, you can hide them or try to fix them, but they are just noise :(

  1. Missing ShapeInfo file! or Missing YAML file!

These are program errors caused by missing arguments, if the executable throws these it probably compiled and linked correctly.