Following the Haiti earthquake, rescue workers, mostly from the United States military, needed to get food and water to the displaced persons. But with destroyed communications, impassable roads, and thousands of square miles, locating the people was critical.
It was known that the people whose homes had been destroyed by the earthquake were creating temporary shelters using blue tarps, and these blue tarps would be good indicators for locating the people—if only they could be located in time, out of the thousands of images that would be collected every day.
The solution is to develop a machine learning model, which could search the images far faster and more thoroughly than humanly possible. The goal is to find the best algorithm that could search the images and locate displaced persons in time for the locations to be communicated back to the rescue workers.