silasxue/bnf_globalization

BNF Globalization Code (CVPR 2016)

Boundary Neural Fields Globalization

This is the code for Boundary Neural Fields globalization method. The technical report of the method can be found at http://arxiv.org/pdf/1511.02674v1.pdf

If you use this software please cite our CVPR 2016 paper:

@InProceedings{gberta_2016_CVPR,
author = {Gedas Bertasius and Jianbo Shi and Lorenzo Torresani},
title = {Semantic Segmentation with Boundary Neural Fields},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2016}
}

Installation

1. VL Feat

   Compile the VL Feat library in the folder 'libs/'

2. Normalized Cuts

   Compile Normalized Cuts code in the directory 'Ncuts_9/''

Usage

To use BNF method with the boundary based affinities, check out the files 'BNF_binary_class_edge_affinity_demo.m' and 'BNF_multi_class_edge_affinity_demo.m' for binary and multi class segmentations respectively. If the boundaries are not available, you can use the demos 'BNF_binary_class_RGB_affinity_demo.m' and 'BNF_multi_class_RGB_affinity_demo.m', which perform our globalization technique using the RGB color affinities.

Notes

This version of the code is slightly different than the one presented in the technical report.