NEC develops heterogeneous object recognition, an image recognition technology for retail goods, including fresh produce and packaged products
- Increases the efficiency of product scanning during payment -
Tokyo, March 5, 2018 - NEC Corporation (NEC; TSE: 6701) today announced the development of heterogeneous object recognition, an image recognition technology available for retail goods, including fresh products, daily delivered foods, and packaged products, as part of increasing the efficiency of product scanning for payment in retail shops, such as supermarkets and convenience stores.
In recent years, there is a growing need in the retail industry to save on labor and achieve fully automated payment systems that capitalize on image recognition technology.
By combining deep learning technology and feature point matching technology, this technology meets this need with highly accurate recognition of a variety of retail goods, such as fresh produce where the appearance of individual pieces differs significantly, to industrial products, such as packaged products that come in a wide variety of remarkably similar designs. Even if the placement of multiple products is poorly organized or scattered, the technology reliably recognizes each product individually.
"With this technology, it is possible to significantly increase the efficiency of product scanning during payment because products can be collectively recognized by simply placing them on the cash register counter, with no barcode or RFID scanning," said Akio Yamada, general manager, Data Science Laboratories, NEC Corporation.
Features of the new technology include the following:
- Accurately recognizes vastly different retail goods, from natural produce to industrial products
Both deep learning technology and feature point matching technology are applied in order to accurately recognize each product. A wide variety of products, including natural products, such as fruits and vegetables where the appearance of individual pieces differs significantly, industrial products, such as packaged products of beverages, snacks, and sundries that come in a wide variety with remarkably similar designs, and daily delivered foods such as pre-prepared meals and delicatessen foods that have the characteristics of both natural products (ingredients/content) and industrial products (product labels). By appropriately combining deep learning technology and feature point matching technology in consideration of these product characteristics, recognition accuracy for a variety of retail goods has been increased compared to that obtained by applying only one of the technologies.
- Reliably recognizes each product, even in a complicated environment where multiple products are scattered or poorly placed
To recognize products in a complicated environment in which many products are poorly placed or scattered, a large number of images of complicated environments are automatically synthesized from images of products captured individually. By using the synthesized images for training in machine learning, the reliability of recognition in a complicated environment has been greatly improved. This enables product recognition accuracy to be increased in a complicated environment in which multiple products are scattered, even with a small quantity of captured image data, without capturing or preparing a large quantity of training images, which is a challenge that is inherent in recognition by machine learning.
For More Information
Heterogeneous object recognition technology
About NEC Corporation
NEC Corporation is a leader in the integration of IT and network technologies that benefit businesses and people around the world. The NEC Group globally provides "Solutions for Society" that promote the safety, security efficiency and fairness of society. Under the company's corporate message of "Orchestrating a brighter world," NEC aims to help solve a wide range of challenging issues and to create new social value for the changing world of tomorrow. For more information, visit NEC at https://www.nec.com.
NEC is a registered trademark of NEC Corporation. All Rights Reserved. Other product or service marks mentioned herein are the trademarks of their respective owners. © NEC Corporation.