wangerniuniu / saliencysegment Goto Github PK

Implementation code (V3) for cvpr 2015 and PAMI 2018:
===========================================================================
W. Wang, J. Shen, R. Yang, and F. Porikli, 
Saliency-aware Video Object Segmentation, 
IEEE Trans. on Pattern Analysis and Machine Intelligence, 40(1):20-33, 2018
========================================================================

The code is free to use for research purposes. If you use this software for research 
purposes, you should cite [1,2] in any resulting publication.

[1] W. Wang, J. Shen, F. Porikli, Saliency-aware geodesic video object segmentation, 
IEEE CVPR, pp. 3395-3402, 2015

[2] W. Wang, J. Shen, R. Yang, and F. Porikli, Saliency-aware Video Object Segmentation, 
IEEE Trans. on Pattern Analysis and Machine Intelligence, 40(1):20-33, 2018

The code has been tested on 32-bit/64-bit Window and 64-bit Linux platforms with matlab 
R2009b/R2011b/R2012b/R2013a/R2015. Please compile  '*.cpp' files in 'subCode'.

===========================================================================
Note:
===========================================================================
The saliency demo code (computeSaliency.m) produces slightly different results 
to those reported in the paper [1]. This is due to changes in the parameters and 
the final saliency results are stored in data/outputs/(videoname)/final_saliency.
Saliency results stored in the folder 'intra-frame_saliency' and 'inter-frame saliency'
are generated by our intra-frame graph and inter-frame graph respectively.  
We directly modify Canny as our edge detection method. 

And use our function 'computeBoundary.m' for computing static boundary 
for better performance with more computation time

===========================================================================
Changelog

===========================================================================
v3: Remove some unuseful functions and fix small bugs
v2: Provide binaries for 64-bit Linux and fix small bugs
v1: First public release 

===========================================================================
Contact Information
===========================================================================

Email:
    [email protected]
	[email protected]  [email protected]

This code uses some public source codes from internet.
We thank all the authors for the release of their codes.