The United States Department of Defence (DoD) is seeking machine learning experts to create computer vision algorithms that can speed up analyses of aerial and satellite imagery.
Hosted by the DoD’s Defence Innovation Unit (DIU), the xView2 Challenge seeks to automate post-disaster damage assessment with “computer vision algorithms that will speed up analysis of satellite and aerial imagery by localizing and categorizing various types of building damage caused by natural disasters”.
In a call for submissions posted on the DoD website on 15 August, the agency said this is the DIU’s second prize competition focused on furthering innovation in computer vision for humanitarian assistance and disaster relief efforts. It runs from August through November.
The contest builds upon the agency’s xView1 Challenge, which sought out computer vision algorithms to locate and identify distinct objects on the ground useful to first responders.
The challenge will be underpinned by a new annotated building damage dataset – xBD – created by a team of academics and industry experts led by the DUI to “enable localization and damage assessment before and after disasters”.
“While several open datasets for object detection from satellite imagery already exist — for example, SpaceNet and xView — each represent only a single snapshot in time and lack information about the type and severity of damage following a disaster”, the DUI noted.
The new database is expected to allow machine learning/artificial intelligence practitioners to generate and test models to help automate building damage assessment.
It uses open source electro-optical imagery encompassing 700,000 building annotations across 5,000 square kilometres in 15 countries, and includes seven disaster types – wildfire, landslides, dam collapses, volcanic eruptions, earthquakes/tsunamis and wind and flooding damage.
The findings will be applied to a number of different operational and academic use cases covering areas such as obstructed roads, resource allocation decision-making, and object recognition and identification among others.
Baseline models, developed collaboratively between DIU and Carnegie Mellon’s Software Engineering Institute, will be publicly available as a starting point for the Challenge.
In addition to advancements in damage assessment, the DUI envisions that the xBD dataset will provide researchers, companies and other groups with the “means and motive to develop algorithms that bring humanitarian assistance and disaster response” into the age of artificial intelligence.
There are three competition prize tracks. The open source track will see teams compete for leader board positions and awards for top scores. By releasing their models publicly under a permissive open-source license, teams become eligible for an additional award.
Teams can join the nonexclusive government purpose rights track when they grant government purpose rights and their solutions will be used to help future disaster recovery efforts.
Meanwhile, teams on the evaluation only track retain their intellectual property and only grant DIU the right to benchmark their solution and compete for leader board positions. Top teams in this last category will still be eligible for a special monetary prize pool for their submissions.
The best solutions for all three categories will be eligible for a share of a US$150,000 prize purse and top solvers will also be invited to present their work at the December NeurIPS 2019 Workshop on AI for humanitarian assistance and disaster relief. Winners of any cash prize will be considered eligible to be awarded follow-on work with the DoD.
“DIU’s goal in hosting this challenge is to enlist the global community of machine learning experts to tackle a critically hard problem: detecting key objects in overhead imagery in context and assessing damage in a disaster situation,” Mike Kaul, DIU artificial intelligence portfolio director, said in a statement.
“We are always looking for ways to improve rapid damage assessment to ensure we and our partners deliver the right resources to the right places at the right time,” the Federal Emergency Management Agency’s (a partner in the challenge) Regional Administrator, Robert Fenton, added. “We are confident the DIU Challenge can contribute to that goal.”
Other partners include NASA Earth Science Disasters Program, , California Governor’s Office of Emergency Services, Cal Fire, the California National Guard, DOD’s Joint Artificial Intelligence Center, Carnegie Mellon’s Software Engineering Institute, the United States Geological Service, the National Geospatial-Intelligence Agency and the National Security Innovation Network.