Skip to content

Home

Jump to bottom
ex-ratt edited this page Oct 6, 2016 · 15 revisions

Welcome to the FeatureDetection wiki!

Dependencies

Required dependencies

  • CMake
  • Boost (including system and filesystem)
  • OpenCV 2 (including nonfree)
  • Eigen 3

Optional dependencies

Some features need Qt5, HDF5, BLAS, Eigen, the Matlab Compiler Runtime or the MS Kinect SDK. They can be enabled on demand in initial_cache.cmake. They are not required and you only need them for special functionality.

Installation instructions

Linux

  1. Create a directory for the whole project, e.g. cd ~ && mkdir ffd && cd ffd
  2. Checkout the code: git clone git://github.com/elador/FeatureDetection.git. It will create a directory ~/ffd/FeatureDetection/ with all the source files. This is your source-directory.
  3. Copy ~/ffd/FeatureDetection/initial_cache.cmake.template to ~/ffd/FeatureDetection/initial_cache.cmake. If you run into problems while running CMake, you can specify installation paths to the required dependencies here.
  4. Make sure you have all the required dependencies: For Ubuntu 13.10, you would do this: sudo apt-get install ...
  5. mkdir ~/ffd/build && cd ~/ffd/build. This is your build-directory. We will build the projects in here.
  6. Run cmake -C ../FeatureDetection/initial_cache.cmake -D CMAKE_BUILD_TYPE=Release ../FeatureDetection. A Makefile will be created in this directory.
  7. Type make to build everything.
  8. Run e.g. ~/ffd/build/ffpDetectApp/ffpDetectApp or ~/ffd/build/faceTrackingApp/faceTrackingApp -c 0.

Note: In the future, there will be an install command that installs the built libs and programs to some dir like ~/ffd/bin/. This is a TODO.

Windows

  1. Install git from http://windows.github.com/.
  2. Open a command-line (cmd.exe), go to e.g. D:\ and create a dir for the whole project, e.g. cd D:\, mkdir ffd, cd ffd
  3. Checkout the code: Open GitHub and check-out into D:\ffd. It will create a directory D:/ffd/FeatureDetection/ with all the source files. This is your source-directory.
  4. Copy ~/ffd/FeatureDetection/initial_cache.cmake.template to ~/ffd/FeatureDetection/initial_cache.cmake. If you run into problems while running CMake, you can specify installation paths to the required dependencies here.
  5. Download OpenCV and Boost and unzip them (see the links below). Edit initial_cache.cmake and follow the instructions for OpenCV and Boost.
  6. mkdir build, cd build. This is your build-directory. We will build the projects in here. You should be in D:\ffd\build now.
  7. Run cmake -C ..\FeatureDetection\initial_cache.cmake ..\FeatureDetection. A VS solution file will be created in this directory. If some library can't be found, edit D:/ffd/FeatureDetection/initial_cache.cmake.
  8. Run cmake --build . or open the Visual Studio solution file that has been created in D:/ffd/build/FeatureDetection.sln and compile in VisualStudio!
  9. Run e.g. ffpDetectApp.

Installing of the windows dependencies

CMake: Install cmake from here: http://www.cmake.org/cmake/resources/software.html.

Boost: Download the newest boost binaries here: sourceforge.net/projects/boost/files/boost-binaries/ Choose the one that ends with msvc-12.0-64.exe.

OpenCV: Download the latest OpenCV binaries on http://opencv.org/.

Random notes

git-cola is a nice git-GUI for linux.

Supported Operating Systems and Compilers

Windows: Visual Studio 2013

Linux: Ubuntu 13.04, 13.10, Fedora 18, openSUSE 13.1

The software may very likely run on other operating systems, however, some manual configuration or installation of dependencies might be required.

Clone this wiki locally