Fujifilm System Services Begins Proof-of-Concept Experiment Using Image Recognition AI and Sales Data to Support Shelf Display Decisions at NewDays Stores

Fujifilm System Services Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; President and Representative Director: Isao Takamura), with the cooperation of JR East Cross Station Co., Ltd., today began a proof-of-concept experiment at one of its NewDays convenience store chain locations in Tokyo. The experiment uses image recognition AI technology and sales data to visualize sales trends of products displayed on shelves, supporting display decisions. In this demonstration, AI identifies products by photographing shelves with a tablet device, and visualizes analysis results linked with sales data. By verifying the effectiveness of using these results for display decisions, the company aims to create sales floors that do not depend on staff experience.

Figure 1-A: Photographing beverage shelves with a tablet Figure 1-B: Visualizing sales trends for each product

Background

In recent years, due to labor shortages and diversification of employment patterns, it has become difficult to rely on experienced staff for store floor layout as before. In store operations, reviewing shelf displays and product assortments directly impacts sales, requiring quick and accurate decisions based on sales trends. However, such display decisions are easily influenced by differences in staff experience and skill levels, leading to operational dependency and uneven workloads. There is a need for systems that allow diverse personnel to handle store floor creation.

Overview of the Experiment

This experiment targets beverage and confectionery shelves, which have a large number of products and frequent changes. It links with regularly updated sales data to visualize sales trends for each product, verifying the effectiveness of a system that supports display decisions.

By photographing shelves with a tablet, a proprietary image recognition AI detects products and links them with individual sales data. It analyzes top-selling and underperforming products and visually displays results on screen. This allows intuitive understanding of sales trends and easier decisions on product placement and replacement.

Figure 2: Image from shelf photography to product detection and sales trend visualization

Image recognition on actual store shelves is considered a high-difficulty area due to product shape and package similarities, as well as variations in product visibility depending on display conditions. In this experiment, the company combines product information and store operation know-how cultivated through new product introduction and store communication services provided to retailers, with image recognition AI technology to achieve high-precision product identification. By using training data that artificially reproduces store shelves and AI technology that can identify new products without additional learning (*1), the system can operate in actual stores with diverse display patterns.

*1: By creating data simulating actual shelf conditions using proprietary technology and using it for training, the system detects products regardless of different display situations per store. Additionally, by adopting a method based on product features, it can identify new products against a database without additional learning.

Experiment Period

June 8, 2026 to September 30, 2026 (approx. 4 months)

Future Outlook

Based on the results, the company will consider adding functions to support product assortment and display position changes. With a view toward multiple store deployment, it aims to commercialize a service that contributes to store operation efficiency and sales enhancement.

The company will continue to leverage the latest AI and IT technologies to solve various store operation challenges and support better store floor creation.

About Fujifilm System Services

As a wholly-owned subsidiary of Fujifilm Business Innovation, the company provides industry- and business-specific BPO services to local governments and companies nationwide. Particularly, it boasts overwhelming market share in systems such as the 'Comprehensive Family Registry System Bookless' and 'Convenience Store Certificate Issuance System' for local governments and public institutions, and is highly evaluated for its end-to-end service from introduction to support. By enabling better communication for local governments and companies, the company aims for truly enriched lives for residents and consumers as final service recipients, under the vision of 'Information services for a prosperous tomorrow.'

https://www.fujifilm.com/fbss/

* Services and product names mentioned in the press release are registered trademarks or trademarks of their respective companies.