With the help of artificial intelligence (AI), farmers may soon find it easier to remove strawberry runners. It could save time and labor, new University of Florida research shows.
Growers need all the help they can get to reduce production costs for an industry estimated at $500 million a year in Florida.
Strawberry plants can grow special stems called runners. These runners grow out from the main plant and make little baby plants – also known as “daughter plants.”
Strawberry nurseries use runners to grow lots of new plants for each season, and it’s an easy way to multiply plants.
But when a field plant grows runners, it uses up energy that could have gone to making bigger and sweeter strawberries. Too many runners can mean fewer and smaller fruits, which can decrease the harvest. Additionally, runners grow into the spaces between plant rows, making it harder for people and machines to pick the fruit.
“Most strawberry growers have to cut runners by hand, which takes time and money. To make this process faster and cheaper, we need machines that can find and remove runners automatically,” said Kevin Wang, an assistant professor of agricultural and biological engineering at the UF Institute of Food and Agricultural Sciences (UF/IFAS).
That’s why the Smart Agriculture Lab and the Phenomics Lab at the Gulf Coast Research and Education Center (GCREC) jointly initiated this project.
Xue Zhou, senior research associate in the Phenomics Lab, trained an AI model to analyze the images and accurately detect strawberry runners.
The AI model learned to recognize runners, even when they looked different — short or long, tangled with leaves or growing at odd angles, said Wang, a faculty member at the GCREC.
“This is the first step toward building machines that can remove runners without human labor, helping farmers grow strawberries more efficiently,” he said. “We found that the AI model worked best when it was trained on a mix of images from different heights and angles. This made it ‘smarter,’ so it could recognize runners in many situations, not just in perfect close-up photos.”
Scientists trained the current AI model using images from strawberry breeding trials. The next step is to collect images from commercial farms to help the model “learn” and adapt to real grower field conditions, Wang said.