A custom of the Jewish New Year is to dip a slice of apple in honey and to wish others a sweet New Year. Symbolizing New Year, a new cycle of reading the Torah and the creation of the world, the apple is high in Technion consciousness right now.

From vision to application, Technion recently made science headlines with an algorithm for identifying apples on trees in order to produce the most accurate count yet for the  future automated harvests.

The estimation of how much fruit is on the tree or on the ground is a challenge for computers as leaves and branches interfere. Agricultural calculations also have to take into account different colours, depending on ripeness, variety, the weather and the time of day.

Raphael Linker and Eliyahu Kelman of the Faculty of Civil and Environmental Engineering have composed an innovative  program to search photographs of lit-up apple trees for glints in the foliage: light reflecting off the shiny fruit.

In one experiment, when humans counted 6713 apples in the pictures, the computer counted 6687 – a vast improvement on automated estimations (Computers and Electronics in Agriculture, doi.org/4hb).

“With farms getting larger, farmers only have the time to look at a few trees,” Linker said. “If you can have an automated system driving over the orchard, you get a much more reliable picture.”