Detecting Change With Artificial Intelligence

摘要

In a project for the Defense Department's Defense Innovation Unit (DIU), computer scientists have turned to artificial intelligence and aerial imagery to construct a detailed damage assessment solution. The tool can be used remotely and automatically to determine the amount of damage to buildings and structures from a natural disaster or catastrophe. The prototype, known as the xView Ⅱ model, was tested this fall, with the goal of rolling out a more finalized operational version next year. In the last few years, the U.S. military has seen an enormous amount of weather-related damage to some of its facilities, including the destruction at Tyndall Air Force Base, Florida, from Hurricane Michael in 2018; extensive water damage at Camp Lejeune, North Carolina, from Hurricane Florences torrential rains in 2018; and flooding of the Missouri River and area creeks that impacted one-third of Offutt Air Force Base, Nebraska, in 2019. Meanwhile, this fall, California's wildfires raged over 4 million acres causing irreparable damage, while repeated hurricanes barraged the Gulf Coast.