Sign In

Communications of the ACM

ACM TechNews

UW Researchers Hone Computer Models to Identify Animals in Photos

View as: Print Mobile App Share:
Image of a mountain lion captured by a camera trap.

University of Wyoming researchers taught computer models to identify images of wild animals by training them on camera-trap photographs.

Credit: Jim Beasley

Computer models developed by researchers at the University of Wyoming (UW) use artificial intelligence to identify images of wild animals in camera-trap photographs.

The models were created using 3 million camera-trap images from 18 studies in 10 U.S. states.

The "species model" recognizes 58 species, while the "empty-animal model" filters out images that do not contain animals; both were 97% accurate when tested on images from areas used in developing the software.

Accuracy rates ranged from 90% to 94% for the empty-animal model and 65% to 93% for the species model when tested with images from other parts of the world.

Said UW's Mikey Tabak, "The poor performance of the species model in some areas indicates that some users will need to train new models on images from their field sites. But the empty-animal model appears to be broadly applicable for sorting out empty images in datasets globally."

From University of Wyoming
View Full Article


Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account