Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1.4 KB

README.md

File metadata and controls

35 lines (22 loc) · 1.4 KB

AF-graph

Source code for the paper "Affinity Fusion Graph-based Framework for Natural Image Segmentation". The AF-Graph paper can be found here.

AF-Graph is modified from the offcial GL-Graph implementation. A list of papers and datasets about natural/color image segmentation can be found in our repository.

Requirements

The code requires the version of Matlab2018a, Ubuntu 16.04.

Data

The BSDS300 dataset and predata (extracted features) can be downloaded from GL-Graph and place them in BSD/ and bsd_300_feat/ folder, respectively.

Demo

Run the demo demo_AF_Graph_BSD300.m in parallel on 4 workers (parfor in KSC_graph).

Results of BSD300

The detailed results can be found in evaluation.txt.

Other results

Results and codes of MSRC dataset can be found in here.

Citing

If you find this repository useful in your research, please consider citing:

@ARTICLE{AF-Graph,  
  author={Y. {Zhang} and M. {Liu} and J. {He} and F. {Pan} and Y. {Guo}},  
  booktitle={IEEE Transactions on Multimedia},   
  title={Affinity Fusion Graph-based Framework for Natural Image Segmentation},   
  year={2021}}