Skip to content

Latest commit

 

History

History
30 lines (18 loc) · 736 Bytes

File metadata and controls

30 lines (18 loc) · 736 Bytes

svm-GraphCut

Interactive user marked segmentation with GraphCuts.

Details

  1. Segmentation algorithm uses Support Vector Machine and Boykov GraphCut to grow user marked pixels into foreground and background regions.

  2. Available in MATLAB and Python.

  • MATLAB requires CVX and Maxflow library
  • Python requires sklearn and PyMaxflow
  1. Interactive GUI for analysis.

Results

  • Green: User marked background.
  • Blue: User marked foreground

church church

fish fish

zebra church

Note: Marking with Python GUI

TODO:

[] Jaccard Similarity on BSD. [] Clean.