Progressively Growing Generative Adversarial Networks for High Resolution Semantic Segmentation of Satellite Images

TL;DR

This paper explores a progressive training approach for GANs to improve high-resolution satellite image segmentation, specifically for rooftops, achieving higher accuracy than traditional methods.

Contribution

It introduces a progressive GAN training methodology tailored for high-resolution satellite image segmentation, demonstrating improved accuracy over conventional GANs.

Findings

Progressive GAN training achieved 93% accuracy.

Traditional GAN training achieved 89% accuracy.

Method outperforms conventional approaches in rooftop segmentation.

Abstract

Machine learning has proven to be useful in classification and segmentation of images. In this paper, we evaluate a training methodology for pixel-wise segmentation on high resolution satellite images using progressive growing of generative adversarial networks. We apply our model to segmenting building rooftops and compare these results to conventional methods for rooftop segmentation. We present our findings using the SpaceNet version 2 dataset. Progressive GAN training achieved a test accuracy of 93% compared to 89% for traditional GAN training.