Intel and the American Red Cross to use artificial intelligence technology for disaster preparedness

The American Red Cross and its Missing Maps project are working with Intel to apply artificial intelligence to map disaster-prone areas in developing countries to help them prepare for disasters. In 2019, Intel data scientists built a computer vision Model that was able to identify previously unmarked bridges and roads on Uganda satellite imagery.

“Accurate geographic information is extremely important for Red Cross staff during disaster planning and emergency response. However, there are parts of the world that are not yet marked on the map, which creates a huge impact on disaster planning and response. That’s why we’re working with Intel to map disaster-prone areas, marking roads, bridges, buildings and cities.” —Dale Kunce, co-founder of the Missing Maps project , CEO of the American Red Cross Cascades Region

According to the Missing Maps Project , nearly 200 million people around the world are affected by disasters every year. Many disaster areas are not marked on the map, leaving emergency responders without the necessary information to quickly make disaster response decisions.

Satellite imagery is sometimes difficult to identify, and bridges and infrastructure vary from country to country. AI models enhance mapping capabilities to cover wider areas and capture things that are imperceptible to the human eye. For example, the model found 70 bridges in southern Uganda that were not found in OpenStreetMap or the official Uganda Bureau of Statistics map.

The computer vision model runs on second-generation Intel® Xeon® Scalable processors with built-in Intel® Deep Learning Boost Technology (DL Boost) and the nGraph compiler.

Although Intel does not own the full rights to the dataset, it is pursuing the opportunity to make the dataset available as an open source resource to researchers and geospatial professionals. In addition, Intel will host workshops on how satellite imagery and AI technology can be used in humanitarian practice, making the most of the datasets and codebases developed for the project.