Categories: GIS Learning

Using Machine Learning and Satellite Imagery for Street Address Generation

Researchers from Facebook and MIT Labs have proposed a new methodology that uses machine learning and satellite imagery to generate street addresses in areas of the world where individual buildings don’t have a unique address.  The methodology divides the  street addressing into two processes.  The first process is segmentation.  During segmentation, road pixels are identified using deep learning from 0.5 meter resolution satellite images.  The second part of segmentation involves developing the road network from these identified pixels.  Next, the road network is divided into regions.  The second process is called labeling. During this process, the regions, road segments, and place markers are named and block letters are assigned to each unit.  The regions are divided into quadrants (N, S, E, W) with the city centered defined as the densest area.  The streets are numbered and lettered based on their proximity and orientation from the centre of the city.  In comparing the results of the model with manually labeled GIS data, the researchers found that the model was able to learn on average 80% of the roads per city.

Satellite image, extracted roads, labeled regions and roads, and meter markers and blocks of an example developing cities. Source: Demir & Raskar, 2018

There have been other efforts to develop an addressing system for regions of the world that don’t have street addressing. What3Words is one such endeavor which assigns a random combination of three words on a grid system of the world that has been divided into 3 meter by 3 meter cells. What3Words has been adopted by the postal system in Mongolia to help with mail delivery.

The study:

Demir, I., & Raskar, R. (2018). Addressing the Invisible: Street Address Generation for Developing Countries with Deep Learning. arXiv preprint arXiv:1811.07769.



Four billion people lack an address. Machine learning could change that. MIT Review

See Also

Published by

Recent Posts

Mapping Where Norway is Moving

Norway is using Copernicus Sentinel-1 data to detect ground movement across the country.

6 days ago

How To Use the Equal Earth Projection With QGIS on the Mac

If you've been eagerly awaiting the ability to try out the Equal Earth map projection in QGIS, here are a…

2 weeks ago

Mapping Coral Reefs

Scientists have been developing better ways to fuse data from different satellite systems and data capture, while also developing classification…

2 weeks ago

PACE: NASA’s Earth Observation Satellite to Launch in 2022

NASA’s PACE (Plankton, Aerosol, Cloud, Ocean Ecosystem) is a planned earth observation satellite with an anticipated launch in 2022.

2 weeks ago

GIS and Map Contests Roundup

Current map and GIS related competitions listed in order of the competition deadline.

4 weeks ago

GEDI: A New Tool for Forest Carbon Maps

Maps created by the GEDI satellite system will enable a detailed measurement of the carbon content of forests, allowing scientists…

4 weeks ago