Blog

08.22.2019

The Rise of the HIVE: A New Era of Crowdsourced Imagery Analysis

By: Aaron Tirrell, GeoHIVE Product Manager, Maxar Technologies


Read Time: 4 minutes

Tomnod opened up the world of satellite imagery analysis to the public nine years ago. This amazing platform gave users the power to make a difference in the world and aid in humanitarian efforts, damage assessment and conservation initiatives. Through Tomnod, the “citizen scientist” community unlocked new insights from Maxar’s satellite imagery that went on to save lives and make the world a better place. As Tomnod winds down, that does not mean the end of crowdsourced imagery analysis at Maxar. In fact, it is evolving.

In 2015, Maxar developed GeoHIVE (Geospatial Human Imagery Verification Effort), a crowdsourcing platform that allows users to discover and verify features of interest in Maxar’s high-resolution satellite imagery to provide insight on even the most complex questions. Today, a dedicated team of Maxar developers and geospatial analysts process and review the crowdsourced results to ensure the highest geospatially accurate data possible. GeoHIVE is unique because it utilizes the collective intelligence of a vetted, compensated crowd for its campaigns. This crowd has grown to over 3,000 members, who have completed more than 700 campaigns, viewed more than 4.5 million square kilometers of the Earth and identified more than 9.6 million features for customers like the U.S. government, the U.S. Department of Defense, the Intelligence Advanced Research Projects Activity (IARPA) and the Canadian Wildlife Service and Parks Canada.

GeoHIVE tailors campaigns to provide specific datasets and solutions for each unique question brought to the team. The platform currently hosts three different types of campaigns:

  • Discovery—Users locate and tag features of interest.
  • Validation—Users confirm the presence or absence of features of interest that have been identified by other sources like machine learning algorithms or volunteered geographic information.
  • Editor—Users draw polygons around features of interest to digitize them.

This is now one of the most powerful platforms in the world for satellite imagery crowdsourcing.

GeoHIVE Infrastructure Discovery Campaign

New Ways to Provide Insight

From humanitarian and conservation efforts to security planning and disaster relief, the GeoHIVE crowd has shed light on some of the most pressing issues currently affecting the world. That said, the GeoHIVE team and crowd cannot do this all on their own.

Artificial Intelligence (AI) and machine learning (ML) are integral capabilities, particularly for removing parts of imagery with extensive cloud coverage and reducing the area the crowd must analyze. This allows the crowd to focus their attention on the best imagery available, ultimately yielding results that are more precise for the customer. The crowd works alongside the Maxar team to teach the computer to identify specific features or imagery patterns by validating features to use in training algorithms. The use of AI/ML allows for faster, more accurate and thorough analysis of imagery, which is particularly essential to campaigns providing information in emergency situations or natural disasters.

Emergency Response to Hurricane Michael

When Hurricane Michael started forming in the Gulf of Mexico, the Maxar team collected the most recent pre-event, satellite imagery around the area of expected landfall. On October 10, 2018, Michael slammed into the Florida Panhandle as a Category 5 hurricane, with winds reaching speeds of 160 mph— the first hurricane of that magnitude to strike the continental United States since 1992.

In the days following the storm, GeoHIVE launched a campaign that utilized the pre-event imagery and post-event Rapid Response drone imagery collected by the U.S. National Oceanic and Atmospheric Administration. The crowd provided a damage assessment of all the buildings affected by the storm. In almost three hours, the crowd viewed 7,167 buildings in the Mexico Beach and Port St. Joe. They labelled 12% of the buildings as visibly destroyed (red dots in the graphic below), 19% as visibly damaged (yellow dots), and 69% as sustaining no visible damage (green dots). If you would like to explore the results, an interactive map is available on the ‘map’ tab as part of our Hurricane Michael - Open Data Program event page.

GeoHIVE Damage Assessment of Mexico Beach, FL.

Thank You, Tomnod

GeoHIVE would not be the platform that it is today without the foundation that Tomnod laid almost a decade ago. Through the Tomnod platform, citizen scientists made a difference in the world. Maxar’s GeoHIVE intends to continue that legacy for a better world.

To learn more about GeoHIVE, please visit our website.

Back to Blog

Email Subscription