N.C. PSI Applies AI to Address Key Ag Challenges

As North Carolina State University prepares to host the national AI in Agriculture Conference, N.C. Plant Sciences Initiative researchers are continuing their work to advance responsible field-ready AI-enabled tools and technology that make a difference for farmers and consumers. 

Convened by N.C. PSI platform director and agricultural analytics expert Daniela Jones (of the Department of Biological and Agricultural Engineering), the conference will provide participants the chance to visit the initiative’s home, the state-of-the-art Plant Sciences Building on the university’s Centennial Campus in Raleigh, and meet N.C. PSI computer scientists, engineers, plant scientists and other researchers behind groundbreaking interdisciplinary projects to solve agricultural challenges with AI.

AI is the defining opportunity of our time in agriculture. If a scientist or farmer can observe it in the field, AI can be trained to replicate and scale that knowledge.

N.C. PSI’s platform director for resilient agriculture, Chris Reberg-Horton, is among them. As the agronomist-turned-software engineer notes, “AI is the defining opportunity of our time in agriculture. If a scientist or farmer can observe it in the field, AI can be trained to replicate and scale that knowledge.

“Just as genomics reshaped plant breeding, digital agriculture and AI are redefining how farming is done,” continues Reberg-Horton (of the Department of Crop and Soil Sciences). “N.C. PSI is figuring out how to make end products out of proof of concepts as it relates to agricultural applications of AI, sensors, robotics and data-driven decision tools moving from the lab to the field in ways that meet farmers’ needs.”

Here are highlights from a few N.C. PSI AI-in-Ag research projects now underway:

Pixels to Power the Future of Farming

Images, and lots of them, are the key to computer vision, a form of AI that’s driven advancements in facial recognition, medical imaging, self-driving cars and more. But the lack of publicly available, high-quality plant images has been a barrier to advancing computer vision in agriculture.

At NC State, a team of researchers and engineers have amassed over 1.5 million high-quality plant images, plus key data about those plants, into a first- and largest-of-its-kind open-source agricultural image repository, or AgIR. Freely available to agricultural researchers, both public and private, the repository could be a game-changer for agriculture, enabling new tools and technologies that make farming more precise and profitable.

The Superhero Robots-in-Training

Grocery shoppers might not think about it, but the bulk of the fresh produce we buy in the supermarket is still tended and picked by hand, and finding labor is one of the toughest challenges faced by U.S. farmers.

With support from the N.C. PSI Makerspace, NC State researchers and students are developing AI-driven robots to automate some of the more arduous and repetitive vegetable production tasks. And they’re naming some of those tools after superheroes.

REFRAME: A Residue Revolution

Not everything that’s grown on a farm makes it to market because it just won’t sell: Misshapen sweetpotatoes and their green tops are good examples. At NC State, researchers have embarked on a four-year project to help create new business opportunities from such agricultural leftovers. They’re developing an AI-enabled, open-source platform to help farmers, processing plants, researchers and policymakers analyze the potential for creating new income sources from such biomass.

Toward Faster Nematode Analysis

To speed test results that farmers need to prevent or reduce losses from nematodes, a team of NC State faculty members are training computers to identify 25 of most pesky types in the Southeast and accurately count them.

Right now, testing is a tedious, time-consuming process, and it can take weeks for farmers to get results. While AI won’t eliminate the need for trained diagnosticians, it could make their challenging work more efficient, allowing farmers to take action when it matters most.

Sweet Success from Sweet-APPS

North Carolina is the undisputed leader in U.S. sweetpotato production, generating over $350 million in annual economic impact and producing roughly 60% of the nation’s supply. NC State’s Sweet-APPS project is designed to ensure the industry stays strong.

Sweet-APPS uses advanced imaging and machine-learning models to evaluate every sweetpotato by size, shape and surface quality, linking that information back to the field it came from. These data are fed back to web-based decision-support tools accessible by packers. The result is smarter grading, less waste and a higher-value crop.

About the N.C. PSI

A rapidly growing population. Less farmland. Climate and water shifts. Emerging crop diseases and pests. The complex challenges of agriculture don’t fit into traditional research silos. The North Carolina Plant Sciences Initiative at NC State University breaks down barriers to solving these and other grand challenges.  We are bringing together the brightest minds in academia, government and industry to drive vital research that increases crop yields, creates new varieties, extends growing seasons, enhances agricultural and environmental sustainability, and produces new and improved technology.