Cucumber sorting tool

A Japanese farmer has used Google’s open-source machine learning library to build a machine that can identify and sort cucumbers automatically.

Wilma den Hartigh
September 15, 2016 10:32 am

In an interview with Google, Makoto Koike said that sorting cucumbers in Japan was time-consuming as each farm had its own classification standard.

The machine transmits data from three cameras to a small computer. The data is then stored in Google’s cloud servers.

The decisions made by the algorithms in the cloud control the machine’s sorting arm, which moves the cucumbers into one of nine groups. An algorithm is a set of instructions used for problem-solving in computer programming.

Before the machine could work, Koike took more than 7 000 images of cucumbers to develop the sorting algorithms. The machine is 70% effective in real world production and can be replicated to sort a variety of crops.

Christiaan Downing, from Unitec SA Fruit Technology Ltd, said that mechanisation was taking place more slowly in South Africa because the cost of labour was lower than in other countries.

“That’s why manual sorting is still widely utilised in South Africa,” he said.

He added that mechanisation enabled greater sorting precision.

“When it comes to fruit, there’s no way a human eye can sort as fast and correctly as technology on the market. Sorting your product is about saving time and money, and being as productive as possible.”