A University of Idaho research team is refining a first-of-its-kind app that will allow grain farmers to instantly identify common pests and beneficial insects inhabiting their fields.??
Marek Borowiecki, with U of I’s Department of Entomology, Plant Pathology and Nematology, is the lead investigator on the project, funded with a four-year, $500,000 Food and Agriculture Cyberinformatics Tools grant offered through USDA’s National Institute of Food and Agriculture.?
Alex McKeeken helped fine-tune the app for his master’s thesis in bioinformatics and computational biology.
The team also includes Sanford Eigenbrode, a professor in the Department of Entomology, Plant Pathology and Nematology; Arash Rashed, a UI Extension entomologist; Jennifer Hinds, with U of I Research Computing and Data Services; Luke Sheneman, program manager of U of I’s Institute of Modeling Collaboration and Innovation; and postdoctoral researcher Subodh Adhikari.
They hope their unique app will be finished and in use in farm fields throughout Idaho by next fall.??
Once it goes live, the app will allow farmers to take photos of insects common in cereals and other Idaho rotation crops with a phone camera to be uploaded and identified.
Borowiecki believes the app will also be of interest to some home gardeners in the state.??
He knows of only one other effort to use artificial intelligence to identify important pests of U.S. crops: an app is in the works to serve Florida citrus farmers. The iNaturalist app can identify many species of plants and animals, but Borowiecki finds it’s not accurate with most insects.??
The U of I research team plans to cover 26 categories of pests and beneficial insects with their app, which should name a species with 85 to 95 percent accuracy.
Growers will be sent links to websites about both harmful and beneficial insects, writeups on the specific pest identified by the app and various Extension resources, such as integrated pest management handbooks for pest species.
The app will also include a social portal to show growers where various pests have been confirmed.
“We have this problem in biology: It’s really hard to identify things,” Borowiecki said. “It often takes a few people to identify something correctly.”?
Idaho is a major producer of several classes of wheat and is the nation’s top barley producing state.
Through promptly and accurately identifying pests, Idaho grain growers stand to prevent crop damage, avoid unnecessary insecticide treatments and understand the ramifications of their treatments on beneficial insects.??
UI Extension currently offers the Idaho Insect Identification program as a free service to growers who submit photos for Extension entomologists to identify.??
“A human has to go through that and figure out what it is,” Borowiecki said. “The hope with this tool is to shortcut that process a little bit – let them know quickly whether they have something to worry about.”?
They need hundreds of photos of each insect covered by their app to improve the accuracy of its identifications and have been working with agronomists throughout the state, who submit photos from the field via a website to bolster their database.
They also invite others to submit photos of Idaho crop pests. Some of the photos in their database were found on the internet.??
As a preliminary test, they added hundreds of images to the database of three common aphid species affecting Idaho grains – bird cherry-oat aphid, cereal grass aphid and English grain aphid – which they took in a greenhouse and at some of the university’s experimental farms.
They trained their algorithm to differentiate between photos of the three species with up to 80 percent accuracy.??
Borowiecki estimates the database is about halfway complete. Larger pests, such as click beetles and wire worms, can be identified with greater accuracy by their program.
The app will identify specimens to the species level if there are important considerations, such as diseases that an individual species may transmit or differences in pest management approaches.
In many cases, the app will identify pests only by category.