Geoseg: A Computer Vision Package for Automatic Building Segmentation and Outline Extraction

TL;DR

Geoseg is a comprehensive software package that unifies multiple deep learning models for building segmentation and outline extraction in remote sensing, emphasizing ease of use, flexibility, and performance evaluation.

Contribution

It introduces a unified, flexible package implementing over nine state-of-the-art models for building segmentation, filling a gap in existing tools.

Findings

All models are evaluated on a high-quality aerial dataset.

The package demonstrates competitive performance and efficiency.

Provides utility scripts for training, evaluation, and visualization.

Abstract

Recently, deep learning algorithms, especially fully convolutional network based methods, are becoming very popular in the field of remote sensing. However, these methods are implemented and evaluated through various datasets and deep learning frameworks. There has not been a package that covers these methods in a unifying manner. In this study, we introduce a computer vision package termed Geoseg that focus on building segmentation and outline extraction. Geoseg implements over nine state-of-the-art models as well as utility scripts needed to conduct model training, logging, evaluating and visualization. The implementation of Geoseg emphasizes unification, simplicity, and flexibility. The performance and computational efficiency of all implemented methods are evaluated by comparison experiment through a unified, high-quality aerial image dataset.