Helping Facebook connect the world with deep learning
Facebook Connectivity Lab is leveraging DigitalGlobe’s Geospatial Big Data (GBDX) initiative
At the Mobile World Congress today, Facebook’s Mark Zuckerberg announced the Telecom Infra Project, an initiative to develop new technologies and approaches for connecting the 4.2 billion people that still remain offline. To enable this, the Facebook Connectivity Lab is leveraging DigitalGlobe’s Geospatial Big Data initiative to determine population densities across vast rural areas in 20 developing nations. If you are going to connect the world, you need to first know where in the world humans live!
Check out Facebook’s white paper, “Connecting the World with Better Maps: Data-Assisted Population Distribution Mapping.”
Existing maps of populations in many parts of the world are too coarse, outdated, and inaccurate. To solve this problem, information from high-resolution satellites proves invaluable; it provides a consistent global information dataset for mapping population locations. This map, in turn, can inform service providers where connectivity infrastructure should be deployed, whether it be fiber networks or wi-fi hotspots, or communication networks with high-altitude balloons or UAVs. The most efficient network technology depends on the ensuring communications networks designed on proximity to population.
This where DigitalGlobe’s content and platform become a critical part in achieving the vision. DigitalGlobe will be completing an accurate mosaic of the globe at 50 cm resolution in the coming months and will be replenishing this basemap of the world on a frequent basis. Further, DigitalGlobe’s Geospatial Big Data platform, GBDX, makes this rich content along with our 15-year digital library of over 90 petabytes of high resolution DigitalGlobe satellite image data available to anyone, for processing in the cloud. Available for use alongside the data are some of our best computer vision algorithms we designed to convert pixels into meaningful data. On the GBDX platform we provide a much richer library by supporting an ecosystem platform which allows experts to bring their own algorithms to the data and share them with others if they so choose.
For Facebook, we combined DigitalGlobe’s GBDX platform with algorithms for mosaicking and atmospheric compensation to provide consistent mosaics of countries that are foundational to extraction population density mapping. Facebook’s convolutional neural nets are using this accurate content to identify populations over a ridiculously large area. And we are just getting started!
By building an ecosystem of algorithms, developers, data partners, and customers, we have a vision of enabling a whole new set of applications that leverage enormous quantities of data along with computation at a scale like never before. As Facebook publishes its results to the community, we aim to make them available on GBDX platform to allow others to build on them and use them to inform their own analysis. Further, we will be placing DigitalGlobe created population estimation tools and feature detection tools at your fingertips for comparison, building on, and innovating.
The industry is at a critical junction where cloud computing and advancements in deep learning are coming together. With accurate content of our planet that is being updated daily and our GBDX platform, we are putting data and our platform at this technological intersection. We can’t do it alone, though. All of you developers, startups, and enterprises are invited to join our GBDX Beta – we can’t wait to see what you will do!
To learn more about GBDX, visit: developer.digitalglobe.com/gbdx/
Dr. Shay Har-Noy is DigitalGlobe’s Vice President & General Manager, Platform