Orbital Insight Shows How Artificial Intelligence (AI) Can Be Used to Study Images from Space

Zachary Romano


“A satellite can cover every square inch of the earth every two weeks,” says Dr. James Crawford, the former NASA employee and widely known for his Google Books project, where he and his team digitized nearly 20 million books. Using a similar process, Crawford has now started his company Orbital Insight.

At present, Google Ventures, Sequioa, and Bloomberg Beta, all venture capitalist groups, have collectively invested $8.7 million into his company with the hope that they can successfully develop analytical tools that will extract useful information from global satellite images.

Using deep learning to analyze geospatial data

In order to analyze such a large set of images, and potential data, Crawford and his company develop “deep learning” processes. Deep learning utilizes a hierarchy of artificial “neurons” to learn and recognize patterns in data. Essentially, the programmers will teach a computer to understand patterns that it can then look for on its own. They are programming the computer’s cognition to be able to independently process and comprehend patterns of information.

Orbital Insight assesses retail sales based on the number of cars in store parking lots. To do so, a human programmer would tag cars in satellite images as the machine observed. Eventually, the computer would know how to spot and identify cars on its own, without a human involved in the process. Machine learning and the development of artificial intelligence have streamlined this process and allows the company to take thousands of images and turn them into workable sources of data.

Free weekly newsletter

Fill out your e-mail address to receive our newsletter!

Orbital Insight uses shadows detected on satellite imagery to track construction rates in Nanjing, China.
Orbital Insight uses shadows detected on satellite imagery to track construction rates in Nanjing, China.

Applications of this analytical tool are theoretically endless. Some examples include the assessment of crude oil containers to see how full they are, which can be done by observing the shadows produced by the top of the containers.

Similarly, new construction projects over time can be tracked with the presence of shadows where they previously did not exist. This proves especially useful when trying to understand markets like China, where information privacy is strictly maintained. Global investors are basing their speculations on the information extracted from satellite imagery with the hopes of informing their investment strategies.

Global non-profits like the World Resources Institute have also partnered with Global Insight to track deforestation. As these types of firms continue to benefit from diminishing costs of satellite imaging technology while further developing deep learning processes, the ability to make informed decisions on a global scale will become an even more realistic endeavor.


Orbital Insight Raised $8.7 Million To Use AI To Analyze Satellite Data – Forbes, March 20, 2015

Startup Promises Business Insights from Satellite Images – MIT Technology Review, March 16, 2015

Photo of author
About the author
Zachary Romano
Zachary Romano is a recent graduate from Brandeis University and an aspiring researcher in urban economics and real estate with a focus on the use of quantitative methods and spatial analysis. He is a recent graduate from Brandeis University where he obtained a B.A. in Economics with a minor in Anthropology. At present, he has committed to a one-year term of service as an AmeriCorps VISTA with the Community Prosperity Initiative in Syracuse. Zach Romano devotes his time to cycling, volunteering with civic organizations, and spending time on the water throughout Central New York. Some of Zach's work: Housing and Transportation Demand Analysis: Boston Metropolitan Area Assessing Transportation Capacity and Property Values In and Around the Boston Metropolitan Area