IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

TroutSpotter App Uses AI to Track Coldwater Fish Populations

The free app that identifies and tracks individual fish will launch in the spring through tu.org, allowing the national nonprofit to put its members to work helping researchers spot trends in coldwater fish populations.

school of fish
(TNS) — Improvement in local streamwater quality is expected to be among the benefits as artificial intelligence drives an app that will make anglers' photos about much more than bragging rights.

The free TroutSpotter app that identifies and tracks individual fish will launch in the spring through the Trout Unlimited website — tu.org — and will allow the national nonprofit to put its thousands of members to work helping researchers spot trends in coldwater fish populations.

Katie Bartling, the watershed project coordinator at Penn State's Agriculture and Environment Center, says she's optimistic that TroutSpotter will provide a deeper connection to nature that will make anglers more curious about their local streams and more attentive to the ecological conditions that make their favorite spots ideal for fish.

"When you start to look at one individual, then you start to build that story of why that individual likes that spot," Bartling says, "you start to get a broader, wider perspective."

Individual anglers will only see their own data, in order to protect the privacy of favorite fishing spots, but researchers will have access to the entire database so they can analyze how particular species are faring in various stretches of water.

As an added bonus, participating anglers will receive a notification when a fish they photographed is spotted again.

"They get a notification saying, 'Hey, you know, we found George,' " says Nathaniel Hitt, a research fish biologist at the U.S. Geological Survey's Eastern Ecological Science Center in West Virginia.

Hitt says he hopes the social aspect of using uploaded photos to track the movements of individual fish will further invest anglers in the health of their local waterways.

TroutSpotter will be a useful tool for anglers of all ages, but Hitt says it has the potential to engage younger fishers who are more interested in technology.

Collaboration builds capacity

The app and the research it will support are a collaboration between the U.S. Geological Survey, Trout Unlimited and Wild Me, a nonprofit that builds the software tools needed to track individual animals through photography and AI-powered computer vision.

There are apps on the market already that can identify animals by species, such as FishVerify or the bird identification app Merlin, but TroutSpotter is a more precise tool.

The app is built off a neural network that is trained to recognize the unique spot patterns on fish such as trout and salmon.

"What we're doing is pushing the boundaries there and taking things to the individual level," Hitt says.

To enable that process, images are digitally sliced into ribbons and then pixels and fed into the model square by square so that the neural network can identify the red, green and blue color values that make up the markings on each individual fish.

Hitt says researchers have found the tool performs at above 90% accuracy identifying fish caught across different seasons and years.

"The technological capacity is there," Hitt says.

The main focus in the eastern regions of the country will be inland trout: brook, brown and rainbow.

Hitt says it's still an open question whether the technology will work for other species, such as smallmouth bass, but he says the technique should work for any fish with spot patterns. Trout have a spot pattern as unique as a fingerprint.

Fueling local conservation efforts

Bartling says the app could be helpful to monitor fisheries for native brook trout.

That species has a narrow temperature window it tolerates, with temperatures above 68 degrees considered stressful for the fish.

Of the wild trout waters identified by the Pennsylvania Department of Environmental Protection, Bartling counted 38 in Lancaster County. She says monitoring those streams could help Trout Unlimited better target its conservation efforts.

As a Trout Unlimited member and a fly angler herself, Bartling says she appreciates all the organization has done to implement stream cleanups and other water-quality projects, and she says TroutSpotter data would prove valuable even if it were only used to better inform the prioritization tools the nonprofit already uses to help select projects.

Bartling says the ultimate impact of the TroutSpotter data could depend on how many of the conservation organizations working to improve Lancaster County's water quality have access to the data.

But Bartling acknowledges that data access could prove complicated — anglers don't want to expose their favorite locations, and conservationists would also prefer those sweet spots remain isolated to avoid putting pressure on the habitat.

Trout Unlimited's Donegal chapter places a strong emphasis on conservation, and its 800 members have become an active force in Lancaster County's water cleanup efforts.

Balancing privacy and public data

Donegal Trout Unlimited membership coordinator Gary Slater says as motivated as the members are to expand conservation efforts, TroutSpotter could be a tough sell if anglers feel the steps taken to anonymize their data are not enough.

Slater says the balance between privacy and data access presents a catch-22: If individual anglers could see the location where a particular fish they caught had been caught again, that would provide useful information that could motivate anglers to use the app; however, providing that information would mean exposing fishing locations, and giving up location data could deter app users.

In other words, any steps taken to increase the data on migratory patterns that individual users can access could compromise the privacy controls designed to encourage app participation.

"Yes, it would be great to know whether these fish have been caught again," Slater says. "But, at the same point in time, nobody wants to give up their location."

Slater says it might take a couple of seasons, or even a couple of years, for anglers to start seeing the benefit of using TroutSpotter.

"I think it could probably work," Slater says. "But you're really going to have to sell it."

Bartling says there are potential solutions to the tricky balance between public access to the data and protecting anglers' privacy: For instance, she says researchers could break down streams into runs of 200 to 500 meters to make any released location data less exact or they could identify a select group of public access fishing spots as pilot case studies and release the data from those locations to show what the app can do.

Trout Unlimited senior scientist Helen Neville says rolling the app out slowly and focusing on a few target locations in the first year will allow Trout Unlimited staff to work out the kinks and ensure the app holds value for both anglers and scientists.

Neville says the ideal situation is to identify locations where app data could be useful to an ongoing research study, so that participating anglers in a specific place are in close touch with scientists and understand what question their data will help answer.

For example, Neville says TroutSpotter data could help identify whether or not fish are using a restored migratory corridor or track population numbers in a habitat area that's undergone conservation work.

© 2024 LNP (Lancaster, Pa.). Distributed by Tribune Content Agency, LLC.