Google AI builds a better cucumber farm
Machine learning helps sort veggies so the farmer can focus on more important work.
Artificial intelligence technology doesn't just have to solve grand challenges. Sometimes, it can tackle decidedly everyday problems -- like, say, improving a cucumber farm. Makoto Koike has built a cucumber sorter that uses Google's TensorFlow machine learning technology to save his farmer parents a lot of work. The system uses a camera-equipped Raspberry Pi 3 to snap photos of the veggies and send the shots to a small TensorFlow neural network, where they're identified as cucumbers. After that, it sends images to a larger network on a Linux server to classify the cucumbers by attributes like color, shape and size. An Arduino Micro uses that info to control the actual sorting, while a Windows PC trains the neural network with images.
It's not a perfect system, at least right now. Koike estimates that it takes about 2-3 days to train the sorting AI, even using very low-resolution (80 x 80) pictures. And even the 7,000 photos Koike used for that training probably weren't enough. While the sorter recognized 95 percent of test images, real-world sorting dipped to about 70 percent. Having said that, it's not the immediate results that matter. The technology can get better, and it hints at a future where robotic farm equipment handles many mundane tasks that previously required a human's watchful eye.
