Machine-learned Regularization and Polygonization of Building Segmentation Masks

TL;DR

This paper introduces a machine learning pipeline that enhances building segmentation masks by regularizing boundaries with GANs and extracting polygonal outlines, improving accuracy and visual quality.

Contribution

It presents a novel combination of FCN, GAN, and CNN to produce regularized, polygonized building outlines from segmentation masks, advancing building footprint extraction methods.

Findings

Accurate building segmentation with realistic, rectilinear boundaries.

Effective polygonization of building outlines for better visualization.

Improved boundary regularization demonstrated across three datasets.

Abstract

We propose a machine learning based approach for automatic regularization and polygonization of building segmentation masks. Taking an image as input, we first predict building segmentation maps exploiting generic fully convolutional network (FCN). A generative adversarial network (GAN) is then involved to perform a regularization of building boundaries to make them more realistic, i.e., having more rectilinear outlines which construct right angles if required. This is achieved through the interplay between the discriminator which gives a probability of input image being true and generator that learns from discriminator's response to create more realistic images. Finally, we train the backbone convolutional neural network (CNN) which is adapted to predict sparse outcomes corresponding to building corners out of regularized building segmentation results. Experiments on three building segmentation datasets demonstrate that the proposed method is not only capable of obtaining accurate results, but also of producing visually pleasing building outlines parameterized as polygons.