An implementation of "Superpixel Segmentation Using Linear Spectral Clustering" on GPU
We provide an executable file in ./bin, source code in ./src and Makefile in ./build This code supports only 64-bit Windows for now, however, you can compile it on Linux using a compiler with C++11 support.
to execute the app, you need to use the following command.
GLSC.exe 118035.jpg 10
- clean the code. some useless code should be removed.
- remove platform dependent stuff.
If you want to compile the code by your self, you have to install OpenCV 2.4.X and Visual Studio 2013. And modify the Makefile in ./build folder the corresponding pathes, and then you can execute the following command in your cmd.
%comspec% /k ""Your\VisualStudio\Path\VC\vcvarsall.bat"" amd64
cd .\build
nmake
We are very happy if you can cite our paper,
@article{GLSC,
author="Ban, Zhihua and Liu, Jianguo and Fouriaux, Jeremy",
title="{GLSC}: {LSC} superpixels at over 130 {FPS}",
journal="Journal of Real-Time Image Processing",
year="2016",
issn="1861-8219",
doi="10.1007/s11554-016-0652-5"
}
The original paper is:
@inproceedings{li2015superpixel,
title={Superpixel segmentation using linear spectral clustering},
author={Li, Zhengqin and Chen, Jiansheng},
booktitle={2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
pages={1356--1363},
year={2015},
organization={IEEE}
}
Note: this source code is under BSD license.
If you have any question about this code, please do not hesitate to contact me at sawpara at 126.com