That day could come in the near future, because Ampatzidis and his research team are starting to perfect a system to detect the potentially deadly Asian citrus psyllid.
Until now, growers have used the so-called “tap sample method” to find and detect psyllids in their citrus trees, Ampatzidis said. Using that method, farmers strike randomly selected branches and count psyllids that fall onto a sheet of paper. That system is reliable and efficient, but labor-intensive and time-consuming, he said.
Through their research, Ampatzidis and his team of scientists found a new system that works accurately and saves time and money.
“We automated the tap sample method, utilizing machine vision and artificial intelligence (AI),” he said. “The system could be a great way to automate scouting procedures in citrus and to be extended to other crop insects.”
Specifically, researchers took their data-gathering gear into a citrus grove at the UF/IFAS Southwest Florida Research and Education Center in Immokalee, Florida. This novel technology consists of a tapping mechanism to strike selected branches and a board with a grid of cameras that take pictures. Then, an AI-based algorithm analyzes the images, detects, counts and finds the falling adult psyllids. Those photos show scientists, and will eventually reveal to growers, if the tree has psyllids.
“The AI-based software can detect and distinguish Asian citrus psyllids from other insects and debris,” said Ampatzidis, a faculty member at the UF/IFAS Southwest Florida REC.
The system includes a GPS device to record the location of each tree and whether the tree has psyllids, he said. The software develops a map with psyllid detections for each scouted tree so scientists can better see the psyllid data.
From our experiments in a grove, we detected psyllids with 90 percent accuracy.
“The data from each tree can be used to generate maps compatible with precision equipment for variable rate application in order to apply the right amount of pesticides only where needed, and hence, decrease agro-chemical use and expenses, and reduce environmental impact.”
The new study is published in the journal Computers and Electronics in Agriculture.