--------------------------------------------------------------------------------
CCS_v1_0
--------------------------------------------------------------------------------
Code written by Se-Ho Lee
Contact: Se-Ho Lee, seholee@mcl.korea.ac.kr

If you want to use this software, please cite:

S.-H. Lee, W.-D. Jang, and C.-S. Kim.
Contour-constrained superpixels for image and video processing.
CVPR 2017.

@inproceedings{SHLEE2017CCS,
 title={Contour-constrained superpixels for image and video processing},
 author={S.-H. Lee and W.-D. Jang and C.-S. Kim},
 booktitle={CVPR},
 year={2017}}

--------------------------------------------------------------------------------
Quick start
--------------------------------------------------------------------------------
This software was developed under 64-bit Windows with Matlab R2016b.
We also successfully run this software under 64-bit Ubuntu 16.04 with Matlab R2017a.
There is no guarantee it will run on other operating systems or Matlab versions.

1) Structured Edge Detection Toolbox V3.0 is required. It can be downloaded at: https://github.com/pdollar/edges/
2) Piotr's Computer Vision Matlab Toolbox is required. It can be downloaded at: https://github.com/pdollar/toolbox/
3) Superpixel generation for both images and videos needs edge detection results from Holistically-nested edge detection (HED) [1].
4) Superpixel generation for videos needs optical flow results from DeepFlow [2].
5) For videos, optical flow results should be saved at input_dir/flow/ with .flo files.
   Also, edge detection results should be saved at input_dir/edge/ with .png files.


--------------------------------------------------------------------------------
LICENSE
--------------------------------------------------------------------------------
This program is released with a research only license.
The license is enclosed along with this program.

--------------------------------------------------------------------------------
REFERENCES
--------------------------------------------------------------------------------

[1] S. Xie and Z. Tu. Holistically-nested edge detection. In ICCV, pages 1395?1403, 2015.
[2] P. Weinzaepfel, J. Revaud, Z. Harchaoui, and C. Schmid. DeepFlow: Large displacement optical flow with deep matching. In ICCV, pages 1385?1392, 2013.