shameempk / fast_seg

Fast Interactive Image Segmentation using Graph-cut.

Dependencies:

1. System wide dependencies
   • Python 3.5+
   • libopencv-dev
   • python3-tk (to show the result window)
2. Project dependencies (Recommending python virtualenv: http://docs.python-guide.org/en/latest/dev/virtualenvs)

   1. All dependencies are enlisted in requirements.txt

      Install them using : pip install -r requirements.txt

Executing the code:

1. Run the main file using python3: python3 fast_seg.py -i <input-image>
   • Will provide a minimal GUI to mark the seed pixels. While marking, switching between "background" and "object" pixels are done using keys 'b' and 'o' respectively. By default GUI initializes in object mode. Object is marked with "red" markings and Background with "blue".
   • Use python3 fast_seg.py -h for help
2. Press ESC after marking the seeds.
3. Output window will provide the results.
4. Output image will be written in running folder, named "out.png"

For any other inquiries file an issue at https://github.com/shameempk/fast_seg .

Research paper:

Research paper can be downloaded from here.

If you find fast_seg useful please cite this paper in your work:

@misc{
naik2019fast, 
title={Fast Interactive SuperpixelBased Image Region Generation}, 
url={https://www.ijitee.org/wp-content/uploads/papers/v8i8/H7423068819.pdf}, 
journal={IJITEE}, 
publisher={International Journal of Innovative Technology and Exploring Engineering}, 
author={Naik, Dinesh and Shameem, Muhammed}, 
year={2019}, 
month={Jun}
}