ABSTRACT
This chapter presents an algorithm for detecting informally developed roads in satellite images, and quantifying the uncertainty associated with the results. Informal unpaved roads are common in developing countries, and arise naturally through human movement, typically in conjunction with informal housing. Informal roads herein refer to any roads created by residents without government approval. The World Health Organisation provides detailed advice on contact tracing which involves, for example, managing and monitoring contacts daily and direct monitoring by a contact tracing team. There is limited research on informal-road extraction because of the unique challenges of informal roads, such as geometric inconsistencies of the width of roads, visually unclear road boundaries and heterogeneous surface reflectance. Uncertainty quantification and accuracy assessment are usually done as separate processes external to the road extraction. A variety of specific methods is aimed at quantifying the uncertainty in remote sensing applications.
