How Artificial Intelligence is Helping Japanese Cucumber Farmers

artificial intelligence japan cucumber farm AI
Makoto Koike, center, with his parents at the family cucumber farm. Google

Two cucumber farmers in Japan have received an unusual boost to their business after their son adapted Google’s powerful artificial intelligence software to carry out the arduous task of sorting vegetables.

Makoto Koike returned to his parent’s cucumber farm in 2015 after working as a computer systems designer in the automobile industry. While helping out on the farm, Koike realized that one of the most time-consuming processes could be overcome through automation.

Taking inspiration from Google’s powerful artificial intelligence computer program AlphaGo, Koike set about designing a sorting system using a $35 Raspberry Pi 3 computer and Google’s open source deep-learning platform TensorFlow.

artificial intelligence japan cucumber farm AI Makoto Koike, center, with his parents at the family cucumber farm. Google

“Each cucumber has different color, shape, quality and freshness,” Koike said. “The sorting work is not an easy task to learn… It takes months to learn the system and you can’t just hire part-time workers during the busiest period.

“When I saw the Google’s AlphaGo, I realized something really serious is happening here. That was the trigger for me to start developing the cucumber sorter with deep learning technology.”

Koike had never used deep-learning artificial intelligence technology but after inputting some images of cucumbers, he realized that TensorFlow could sort the vegetables with a very high degree of accuracy. Once the software had been fine tuned, Koike was able to build a fully-operational sorting machine that was able to automate the process of sorting cucumbers.

“It’s not hyperbole to say that use cases for machine learning and deep learning are only limited by our imaginations,” Kazunori Sato, a developer on Google’s Cloud Platform, wrote in a blogpost.

“Google offers Cloud Machine Learning (Cloud ML), a low-cost cloud platform for training and prediction that dedicates hundreds of cloud servers to training a network with TensorFlow… and you just pay for what you use, making it easier for developers to try out deep learning without making a significant capital investment.”