Global researchers work to build an app to help farmers

Iowa State's Arti Singh is leading a project that builds on the Pest-ID web-based tool that helps farmers across Iowa and around the world identify and manage crop threats. The technology will be like having an "expert crop advisor or extension agent in your pocket," say researchers from four countries working together to build an app and chatbot to help farmers.

A farmer in Iowa, for example, could snap a photo of an insect pest, a weed or disease symptoms and upload it to the app. The app's artificial intelligence system, trained on millions of images and data points, will provide "instant identification and practical advice on how to manage the problem effectively," according to a summary of the project led by Arti Singh, an Iowa State University plant scientist. (See sidebar for other team members.)

The technology would work equally well across the globe, "making advanced pest management accessible to farmers everywhere, from small family farms to large agricultural operations," according to the summary.

Advancing innovations

Creating the app is part of a four-country initiative called AI-ENGAGE, Advancing Innovations for Empowering NextGen AGriculturE.

"By integrating current and emerging technologies, like AI, into agriculture, we are advancing scientific frontiers to provide U.S. farmers and their international counterparts with tools they need to increase crop yields, more effectively manage pests, strengthen agricultural resilience and ensure a more secure food supply," said Brian Stone, the NSF chief of staff, now performing the duties of the agency's director.

Singh—an associate professor of agronomy, a faculty member with the university's Translational AI Center, a member of the management team of the AI Institute for Resilient Agriculture based at Iowa State and a member of the long-running campus "Soynomics" research team—is leading the U.S. effort to develop the BRIDGE app.

The project will be built on a decade of work by Iowa State's Soynomics team to use AI tools to help farmers in Iowa and around the world identify and manage threats to their crops. The current version, Pest-ID, analyzes photos of insects and weeds to make identifications and recommend potential controls.

Images, images, images

The latest effort will be to add disease identification and management to the web-based app.

"Since we started working on the Pest-ID app, the missing part has been disease identification," Singh said. "Our farmers who have been using the Pest-ID app would like to see diseases on the app."

But it takes millions of images to make these applications useful and accurate. Singh said the insect model uses 16 million insect images; the weed model uses 15 million plant images.

"For diseases, we don't have the millions of labeled images," she said.

Research partners in Australia, India and Japan have already been working to build up datasets for crop diseases, Singh said. Those will help form the basis for a "global-to-local" approach that uses international/global datasets and then fine-tunes them for regional and local pest identification.

Building a bridge

The project's name, BRIDGE, communicates some of the research challenges ahead.

"We're trying to bridge the knowledge gap that we have in our global datasets, especially for the crop diseases and underrepresented pests," Singh said. "We want to connect and strengthen these resources so the tools work better locally in the U.S."

Doing so, "can empower our farmers," she said.

As the project's grant title says, the researchers' joint efforts with AI tools are about "bridging global knowledge and local needs" to provide farmers with "enhanced agricultural production, sustainability and resiliency."

Provided by Iowa State University