LANSING, Mich., (WPBN/WGTU) — The Michigan Department of Natural Resources (DNR) is employing cutting-edge technology to better understand the state’s elk population.

This fall, the DNR will sift through hundreds of thousands of images captured by motion-sensing game cameras deployed across the Northern Lower Peninsula.

Tyler R. Petroelje, a DNR wildlife research specialist, said, “We have right around 1,200 elk is our latest estimate for the Northern Lower Peninsula.”

Traditionally, the DNR has relied on aerial sight ability surveys, flying over elk ranges to estimate populations.

“So in recent years, what we’ve done is we’ve used an aerial sight ability survey and what the survey does is we fly transects over elk range,” Petroelje said. “Some of the challenges that we’ve had in recent years is that snow cover has been less reliable and the sightability survey really relies on good snow cover so that we can detect elk readily.”

However, unreliable snow cover has posed challenges in recent years.

“So we’ve started up a camera survey where we’ve deployed just over 200 cameras across that same core elk range in the Northern Lower Peninsula,” Petroelje said.

These cameras, set up in July, will be collected in October for image review.

“So we deploy these cameras in July and we’re going to be picking these cameras up in October to do image review and they are just your same basic trail cameras that you can pick up off the shelf,” Petroelje said.

The camera survey offers advantages, Petroelje explained: “The nice thing about this camera survey is we’re doing it at a time of year where the bulls have full antler development. They’re readily identifiable between bulls and cows. In a still image, you’re able to get that information, collect that information. Then also between cows and calves, we’re doing it at a year at a time of year where calves are really identifiable between adults and young of the year.”

collect that information. Then also between cows and calves, we’re doing it at a year at a time of year where calves are really identifiable between adults and young of the year.”

The DNR anticipates reviewing between 1,000 to 3,000 photos per camera, amounting to hundreds of thousands of images.

To manage this data, the Michigan Cooperative Fish and Wildlife Research Unit at Michigan State University will use artificial intelligence to filter and analyze the photos.

“So one of those tools can filter images by animal, by vehicle, by human and empty and empty photos and photos of people and vehicles that can be, you know, 40 to 60% of your data set,” Petroelje said.

The AI will help identify species in the photos, with human verification ensuring accuracy.

“Then we can actually utilize another AI tool to try to predict what species are in each of those photos that were designated as animal photos. And then we can also use a human in the loop verification so that even though we have AI going and predicting that there’s an elk in the photo, we want to be very confident that that truly is an elk and then to do what’s the sex of that elk, what’s the age class of that elk,” Petroelje said.

The DNR aims to maintain a healthy elk population, with a management goal of 500 to 900 animals in the region.

“We have a relatively narrow population goal in our elk management plan to have between 500 and 900 animals in the northern lower peninsula and so we need a good amount of precision with a population estimate to try to make sure that we can see where we’re in that goal for the elk herd,” Petroelje said.

This is the first year of the widespread camera study, and if successful, it may become a long-term strategy, reducing reliance on aerial methods.

The DNR plans to share the findings once the data is available.