Intelligent self-service scales, currently being tested in European supermarkets, automatically recognize which type of fruit or vegetable has been placed on them. ©Fraunhofer IITB.
A fruit or vegetable placed on the scale is registered by an integrated camera. An image evaluation algorithm compares the captured image with stored data, thereby identifying it.
Though the concept seems simple, its implementation was fraught with difficulties: Store lighting isn’t optimal for imaging purposes, while produce comes in many shapes, sizes and colours within each category and is usually confined in a cloudy plastic bag. To compensate for these problems, the researchers designed the system to be tolerant to fluctuations in colour and brightness, and they ensured that it could be used under a variety of illumination conditions and with various kinds of cameras. They also included a feature that allows store employees to update the system, enabling it to recognize new varieties of produce.
The system is being tested in approximately 300 supermarkets across Europe.