Salient-Building-Outline-Enhancement-and-Extraction-Using-Iterative-L0-Smoothing-and-Line-Enhancing

ICIP 2019 paper

For the original Pasadena building image, please visit http://www.vision.caltech.edu/html-files/archive.html

For the groundtruth annotation, please visit http://www-scf.usc.edu/~choyingw/

In this repo, we provide some samples from Google Maps and their generated mask. We use training data of ADE20K (https://github.com/CSAILVision/sceneparsing) to train networks of semantic segmentation. Also some pretrained models are available here (https://github.com/CSAILVision/sceneparsing)

Please run the main.m for demo.

If you find this work useful, please consider cite our work

@article{wu2019salient,
title={Salient Building Outline Enhancement and Extraction Using Iterative L0 Smoothing and Line Enhancing},
author={Cho-Ying, Wu and Ulrich, Neumann},
booktitle={ICIP}
year={2019},
publisher={IEEE}
}