AI could help reduce wildlife deaths caused by fencing

Researchers are turning to AI to help wildlife managers find remote fencing that impairs wildlife movement or causes wildlife deaths. Around 1 million kilometers of fencing, largely built to contain livestock, stretches across the western U.S. In recent years, efforts to remove or replace fencing have gained traction as species like elk (Cervus canadensis), deer and pronghorn (Antilocapra americana) are unable to cross them. The problem for many conservation organizations, though, is that the fences are hard to find. To better map the potential wildlife barriers, Wenjing Xu, a postdoctoral researcher at the Senchkenberg Biodiversity and Climate Research Center, and Zhongqi Miao, an applied research scientist with the Microsoft AI for Good Lab, trained a computer to pick out fences from aerial images taken in southwestern Wyoming. The team reported that the system was able to identify about 70% of fences when compared to on-the-ground data. The preliminary research demonstrates how wildlife managers can use this technology across more habitats in the future.

Read more in Science.