University Of Wyoming Team Creates Computer Model To Identify Animals In Images

Sep 29, 2020

Camera traps are used to monitor invasive species like the wild pig.
Credit Valentin Panzirsch

University of Wyoming researchers have created a computer model that can identify animals in images taken by camera traps.

Western Ecosystems Technology's data scientist Mikey Tabak said camera traps give ecologists important information without disturbing wildlife.

"Camera traps could be used in all kinds of studies [like] looking for threatened or endangered species," he said. "In general, as ecologists, we do a whole lot of different wildlife studies where we go out in the field, and we look and see what's there."

He said the problem with the cameras is the large number of images.

"When you have a camera trap out there, it's motion-activated, and so you end up getting millions of images and trying to process all of those images can take a really long time," said Tabak. "So what we wanted to do was create a deep learning model that would automatically recognize the species in those camera trap images."

He said the model will make it easier for ecologists to use the data from camera traps all over North America.

Have a question about this story? Please contact the reporter, Ashley Piccone, at apiccone@uwyo.edu