Fruit growers are testing artificial intelligence tools that count crops, assess ripeness, and forecast harvest dates as weather makes planning increasingly difficult. The systems combine images, historical farm data, weather, and irrigation conditions to help growers decide when seasonal workers will be needed and how much fruit they are likely to pick.
Joel Carter of Okanagan Specialty Fruits experienced the problem during last year’s apple harvest in Washington State, when temperatures reached about 38C and workers had to stop picking at 10 a.m. His company manages more than 1,250 acres of orchards producing apples primarily for sliced portions sold to customers including hotels and schools.
“You need to know more than just when your fruit is going to be ripe,” Carter said. “How long do you have to pick it?”
Cameras Track Fruit From Bud to Harvest
Okanagan is experimenting with technology from Canadian company Vivid Machines, which mounts cameras on tractors to record trees as they pass. AI then identifies buds, flowers, and fruit in the footage to estimate crop volumes and forecast harvest dates.
Carter said the system can identify tiny flower buds that are difficult to spot visually. Its accuracy, however, depends heavily on historical information from each farm because yields can vary according to local conditions and growing practices.
The timing becomes more critical for berries. While Granny Smith apples can have a harvest window of about three weeks, strawberries and other berries may need to be picked within days before quality deteriorates.
UK-based FruitCast uses footage from drones, smartphones, or vehicle-mounted cameras to forecast strawberry, raspberry, blackberry, blueberry, and tomato harvests. The company says its forecasts are within 10% of actual picked volume one week ahead and within 17% three weeks ahead.
Researchers Look Deeper Into Fruit Ripeness
Researchers are also developing tools that measure the condition of individual fruit. Princeton University researcher Yasaman Ghasempour and her students have developed a millimetre wave-based ripeness detector that measures signals associated with water, humidity, and sugar deeper inside fruit.
Other projects focus on making crop counting more accessible. North Carolina State University researcher Jing Zhang is working on technology that can count blueberries from smartphone images, while University of Florida researcher Kevin Wang has developed tools that can gather imagery using drones costing about $100.
Growers still face questions about cost, accuracy, and sharing commercially sensitive information such as fertiliser and irrigation strategies. Ben Palone of Western Growers said forecasting technology could serve as an optimisation tool, but growers are still likely to keep people involved in critical harvest decisions.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.