New method detects 83% of immature citrus; helping cut costs

By detecting green, immature citrus more accurately and efficiently, growers can plan when and where to apply nutrients when fruit is growing and estimate their yield and profit before harvest, said Daniel Lee, a UF/IFAS professor of agricultural and biological engineering.

Using a consumer-grade digital camera, Lee and his colleagues calculated color differences between the fruit and non-fruit objects, and identified fruit using a pre-determined fruit template. They also removed any incorrectly detected fruit via a shape analysis, Lee said. In a newly published study, scientists took 126 images of fruit on trees and detected 83 percent immature citrus, using a camera and the new algorithm. This method is different than the previous ones, which can detect fruit from the images taken farther away from the trees.

Detection of mature citrus on a tree is relatively easy because the fruit looks so different from the background in the camera images, Lee said. But the color of immature citrus closely resembles that of the leaves, rendering it more difficult to find.

The finding means growers can count and find the immature citrus faster, Lee said, which could help growers, who rely heavily on manual labor to pick the citrus, save money, the study says.

"If the growers know fruit yield in advance at different locations, they can manage their grove differently to increase yield, quality and profit," Lee said. If there is less fruit at one site, growers can figure out what's causing that and manage the area differently. For instance, growers can apply more nutrients and water or treat a disease, he said.

Growers count on early and accurate yield forecasts of immature green fruit ahead of harvesting. Such forecasts help growers identify site-specific growth conditions of trees at an earlier stage, the study said.

This so-called "yield-mapping" can also help farmers increase crop yield, while minimizing costs by determine the labor needed during the pre-harvesting and harvesting periods, Lee said.

However, Lee cautions that more research is needed on this method of finding immature citrus. For example, he and his colleagues would like to find a way to get real-time yield in the groves.