Research says AI technology to divert invasive species in Illinois river akin to ‘crime scene photos’

(The Center Square) – Researchers are testing a way to use facial recognition technology to selectively divert invasive Asian carp species in the Illinois River.

The Nature Conservancy, Illinois’ Department of Natural Resources and Whoossh Innovations are testing the use of artificial intelligence to weed out Asian carp at Emiquon, a 6,000-acre wetland along the Illinois River, the News Tribune reported.

Assistant Chief of Fisheries for Illinois Department of Natural Resources Kevin Irons says they are combining AI technology with a fish ladder, which is often used out west to help large river fish like salmon climb dams.

“If carp can climb a fish ladder then we can do some things with it,” he said. “We could identify them and pull them out of that stream and we can put them in a bucket, a bin, a net to hold them to be removed.”

Irons describes the sorting process as akin to crime scene photos.

“Using this fish ladder, there’s an opportunity when they get to the top of the ladder the technology basically takes many pictures in just a second,” he said. “From that information, we can tell what species of fish that is.”

The AI uses a “good fish/bad fish” identification to assist with sorting, eliminating much of the physical labor required, Irons said.

Unlike other barriers, this method would also allow native species to pass through.

“This does give us an opportunity to put those deterrents in place for carp and provide a solution for native fish to pass while still being able to screen those fish to make sure we don’t pass these carp unnecessarily,” Irons said. “So it really allows us to manage our rivers for their best integrity even when battling these invasive fish.”

Keeping the invasive fish out of the Great Lakes is a big part of the researchers’ goal.

“Asian carp, in general, compete with all of our native fish for food – they eat the plankton resources that all fish need when they’re small, and they’re very good at it,” he said.

Irons said they are very happy with how the experiment is going.

“We’re getting images of fish as they go across and nearly 100% accurate on what we can allow past and what we’ll pull back.”

Submit a Comment