Harnessing AI with drone imagery to detect costly celery disease before symptoms surface

University of Florida scientists are launching a new, grant-funded research project that uses artificial intelligence and drone technology to help Florida celery growers detect a damaging fungal disease before symptoms are visible — a breakthrough that could save crops, cut costs and reduce fungicide use.

With support from two grants, a UF Institute of Food and Agricultural Sciences (UF/IFAS) Launching Innovative Faculty Teams in AI and a Southern Sustainable Agriculture Research and Education grant, the project targets early blight, a fast-spreading disease that can quickly turn healthy celery fields into costly losses.

Celery, is a high-value winter crop in Florida, supplies fresh produce to northern states and supports the state’s agricultural economy. Diseases such as early blight disease, caused by the fungus species Cercospora, can significantly reduces crop quality and yield.

Led by plant pathologist Katia Viana Xavier at the UF/IFAS Everglades Research and Education Center, the project will use drones equipped with advanced imaging technologies to identify signs of disease in celery fields earlier and more precisely than traditional scouting methods allow. Larissa Carvalho Ferreira, a postdoctoral research associate on the team, implements the drone-assisted monitoring system with real-time weather data to develop an AI-powered tool for targeted fungicide applications.

In Florida, celery is grown on approximately 1,500 acres each year within the Everglades Agricultural Area and the surrounding city of Clewiston, using conventional and organic production systems to supply markets across the country. Florida ranks third nationally in commercial celery production, behind California and Michigan, making early disease detection a priority for growers supplying markets across the country.

Early disease detection is critical in South Florida, where diseases such as early blight can spread quickly and leave growers with limited management options. By identifying problems sooner, researchers say growers can respond more precisely, reducing crop losses, lowering production costs and minimizing unnecessary fungicide applications.

“By the time you can see disease with the naked eye, it’s often already spreading,” said Xavier. “This project is about giving growers better information sooner, so they can make smarter decisions in the field.”

Instead of relying solely on field scouts which is performed by crop consultants and lab tests, the UF/IFAS team will experiment with drones fitted with multispectral cameras that capture visible light and wavelengths invisible to the human eye. The UF/IFAS team will use advanced computational tools, including machine learning models, to analyze images and detect subtle stress signals in celery plants before symptoms appear.

This precision allows growers to treat only affected areas, rather than spraying entire fields as a precaution, saving money while protecting crops and surrounding ecosystems in the Everglades Agricultural Area. 

“Reducing unnecessary fungicide use protects both the farm and the surrounding ecosystem,” Xavier said. “This is about smarter farming, not just more technology.”
While celery is the focus, the drone system will also support research on other crops, including lettuce, and help advance precision agriculture across Florida.

Growers and crop consultants are directly involved in testing the system, including commercial field trails on commercial farms. UF/IFAS Extension specialists will host field days, workshops and hands-on training to ensure farmers can confidently adopt the technology.

The first celery field day will be held at the end of February for growers, crop consultants and Extension professionals and hosted by UF/IFAS Extension Palm Beach County in Belle Glade. A drone flight mission demonstration will be conducted over the celery field and show differences in disease expression across several plots. These plots are needed to teach the software/model to distinguish healthy versus diseased plants and to detect disease-associated signatures.

A user-friendly, web-based platform is also being developed, allowing growers to upload drone images taken from their fields and receive real-time disease risk assessments based on local weather data.

“This is research designed to be built hand-in-hand with growers, ensuring it works in real-world conditions,” said Xavier.