In this paper we present a method for building boundary refinement and
regularization in satellite images using a fully convolutional neural network
trained with a combination of adversarial and regularized losses. Compared to a
pure Mask R-CNN model, the overall algorithm can achieve equivalent performance
in terms of accuracy and completeness. However, unlike Mask R-CNN that produces
irregular footprints, our framework generates regularized and visually pleasing
building boundaries which are beneficial in many applications.
Description
Regularization of Building Boundaries in Satellite Images using Adversarial and Regularized Losses
%0 Generic
%1 zorzi2020regularization
%A Zorzi, Stefano
%A Fraundorfer, Friedrich
%D 2020
%K segmentation
%T Regularization of Building Boundaries in Satellite Images using
Adversarial and Regularized Losses
%U http://arxiv.org/abs/2007.11840
%X In this paper we present a method for building boundary refinement and
regularization in satellite images using a fully convolutional neural network
trained with a combination of adversarial and regularized losses. Compared to a
pure Mask R-CNN model, the overall algorithm can achieve equivalent performance
in terms of accuracy and completeness. However, unlike Mask R-CNN that produces
irregular footprints, our framework generates regularized and visually pleasing
building boundaries which are beneficial in many applications.
@misc{zorzi2020regularization,
abstract = {In this paper we present a method for building boundary refinement and
regularization in satellite images using a fully convolutional neural network
trained with a combination of adversarial and regularized losses. Compared to a
pure Mask R-CNN model, the overall algorithm can achieve equivalent performance
in terms of accuracy and completeness. However, unlike Mask R-CNN that produces
irregular footprints, our framework generates regularized and visually pleasing
building boundaries which are beneficial in many applications.},
added-at = {2022-07-19T17:16:21.000+0200},
author = {Zorzi, Stefano and Fraundorfer, Friedrich},
biburl = {https://www.bibsonomy.org/bibtex/2812662138ab224e291c55cc3868176ba/redtedtezza},
description = {Regularization of Building Boundaries in Satellite Images using Adversarial and Regularized Losses},
interhash = {13af9738488f29a939f9a3567aa84aa8},
intrahash = {812662138ab224e291c55cc3868176ba},
keywords = {segmentation},
note = {cite arxiv:2007.11840},
timestamp = {2022-07-19T17:16:21.000+0200},
title = {Regularization of Building Boundaries in Satellite Images using
Adversarial and Regularized Losses},
url = {http://arxiv.org/abs/2007.11840},
year = 2020
}