American Red Cross

Many regions around the globe that are most susceptible to natural disasters are also missing critical maps that emergency response workers need to get supplies in, and people out. Intel teamed up with the American Red Cross to build a deep learning algorithm that could significantly speed up mapping of these regions using satellite imagery.

Challenge

Preparing for and responding to disasters requires access to accurate maps of affected regions – maps that do not exist in many of the most vulnerable regions around the world. Current mapping methods use time- and resource-intensive manual labeling from teams of volunteers. 

Solution

Using the Red Cross work in Uganda as a starting point, Intel AI Lab, MILA, and CrowdAI developed a deep learning image segmentation model trained to identify bridges, roads, and waterways. Overcoming significant inferencing barriers, the final model successfully extracted 70 previously unmapped bridges in Uganda.

Missing Maps Project

As the use of deep learning-based mapping has only begun, efforts like the Missing Maps Project (founded in part by the American Red Cross) will continue needing volunteers to fill in the blank areas of under-mapped regions. Get involved by joining or organizing a mapathon.

Learn More

2020 CES

At CES 2020 on January 6, Intel CEO Bob Swan announced that Intel worked with the American Red Cross, the Missing Maps project, and MILA to use artificial intelligence to map bridges and associated roads and waterways in Uganda, a country that is regularly affected by natural disasters and that lacks accurate maps.

Read More

Stay Connected


Keep tabs on all the latest news with our monthly newsletter.

Subscribe