Central-eastern Brazil, by Copernicus Sentinel-2A satellite. Contains modified Copernicus Sentinel data, via Wikimedia Commons
DigitalGlobe, one of the world’s biggest providers of satellite imagery, has announced a major expansion of their partnership with geospatial analytics firm Orbital Insight.
The announcement represents a significant expansion of potential for the application of AI and machine learning algorithms to photographic image processing, also known as computer vision.
DigitalGlobe has given Orbital Insight access to 40 petabytes of their satellite imagery, giving their computer vision engineers, artificial intelligence experts, and data scientists carte blanche on almost 80 times the amount of data they could feed their algorithms under the previous agreement.
“We want to let them race,” Shay Har-Noy, DigitalGlobe’s vice president and general manager of platform, told Wired magazine.
Machine learning algorithms become more powerful according to the dimensions of the datasets they are fed, so this drastic expansion of access to satellite imagery will have a profound effect on the insights that can be gleaned through Orbital Insight’s analytic services. As the sheer scope of material that can be analysed increases, object recognition databases created by the algorithms as they ‘learn’ to classify them can grow exponentially, meaning the kinds of analysis and prediction services that can be delivered will mature much more rapidly.
By means of example, Orbital Insight recently leveraged DigitalGlobe imagery to process 4 trillion DigitalGlobe pixels in a 48-hour period, counting 700 million cars that yielded insights into national shopping behavior for one of their Wall Street clients. This fall, Orbital Insight expects to release several new data analytics products, including worldwide monitoring of crude oil storage inventories and monitoring construction rates across the major cities in China, which are powered by DigitalGlobe’s image library and Geospatial Big Data platform (GBDX).
One can imagine the range of new applications that open up with an increase in available data by a factor of 80. This kind of analysis has been made possible by DigitalGlobe’s transition of their 100-petabyte catalogue of satellite imagery to the cloud, via Amazon web services and semi-trailer, in May this year.
“The era of geospatial big data has arrived,” Dr. Shay Har-Noy said on DigitalGlobe’s blog.
“DigitalGlobe’s platform provides the largest commercial library of current and historical satellite imagery available, integrated into a cost-effective environment where meaningful insight can be extracted from the data through algorithms built by companies like Orbital Insight. We are thrilled to partner with Orbital Insight to provide their customers the most advanced imagery and near-real time analytics required to make accurate and actionable business decisions.”
