GrabCut icon indicating copy to clipboard operation
GrabCut copied to clipboard

NOTE: Image segmentation that iteratively uses Expectation Maximization for Gaussian Mixture Model estimation and Graph Cuts.

GrabCut

Image segmentation using GMM EM and Graph Cuts An implementation of the work described here: http://cvg.ethz.ch/teaching/cvl/2012/grabcut-siggraph04.pdf

Getting Started

You can run GrabCutExample data/soldier.png data/soldier_selection.fbmask result.png to run an example segmentation for yourself.

Build notes

This code depends on c++0x/11 additions to the c++ language. For Linux, this means it must be built with the flag gnu++0x (or gnu++11 for gcc >= 4.7).

Dependencies

  • ITK >= 4
  • Boost >= 1.51 You can tell this project's CMake to use a local boost build with: cmake . -DBOOST_ROOT=/home/doriad/build/boost_1_51
  • Eigen >= 3.2 You can tell this project's CMake to use a local Eigen build with: cmake . -DEIGEN3_INCLUDE_DIR=/home/doriad/src/eigen-3.2.1/