Image-Guided In-Store Shopping Support With Candidate Questions

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing AI-based shopping support systems struggle with accuracy when users are unfamiliar with machine operations or inputting questions, leading to inappropriate answers.

Innovation Solution

An information processing device that receives images from user terminals, performs image analysis to identify content, retrieves question history information, generates candidate questions, and provides answer text through a generative model, optionally incorporating merchandise and device information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If users are required to input questions manually to an AI model, then the system can provide automated answers, but users unfamiliar with machine operations or prompt engineering cannot effectively use the system

Engineering Contradiction:
Improveautomated answering serviceVSAvoiduser operation simplicity
Core Design Contradiction:
Extent of automationVSEase of operation

Solution Approach 1:

The patent introduces an intermediary component that captures images of physical items or documents and automatically converts them into structured queries for the AI model. This intermediary layer (image capture device + image analysis processing) mediates between the user's simple action of taking a photo and the complex requirements of the AI question-answering system, eliminating the need for users to manually craft prompts while preserving automated answer generation

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the mechanical interaction of manual text input with an optical system (image capture and analysis). Instead of requiring users to physically type or speak questions, the system uses image capture devices and automated image analysis processing to extract information and generate queries, substituting a complex manual operation with a simpler visual interaction

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If the AI model processes complex or improperly formatted questions, then it may fail to provide accurate answers, but requiring perfectly formatted questions reduces accessibility

Engineering Contradiction:
Improveanswer accuracyVSAvoiduser operation flexibility
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent performs preliminary action by automatically analyzing images and pre-processing the visual information into structured data before it reaches the AI model. The image analysis processing extracts relevant features, identifies objects, and formulates proper queries in advance, ensuring the AI receives well-formatted input without requiring users to manually prepare perfect questions

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements self-service through automated image analysis processing that independently handles the complex task of converting raw image data into AI-ready queries. The system serves itself by automatically understanding the image content, extracting relevant information, and formulating appropriate questions without human intervention, thereby maintaining answer accuracy while accepting diverse user inputs

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4632656A1Information processing device and information processing method for in-store shopping support
Publication Date: 2025.10.15 TOSHIBA TEC KK
  • EP4632656A1 patent drawingFigure 1~2
  • EP4632656A1 patent drawingFigure 3~4
  • EP4632656A1 patent drawingFigure 5~6

AI summary

According to one embodiment, an information processing device for in-store customer shopping support has a processor connected to a communication interface and a storage unit. The processor receives image information corresponding to an image from a user terminal, executes image analysis processing to identify the content of the image, then acquires question history information correlated to the identified content that corresponds to previous user questions associated with the content in the image. The processor generates candidate questions from the acquired question history information, causes the generated candidate questions to be displayed on the user terminal in a selectable state, then acquires answer text corresponding to a candidate question selected by the user by inputting the selected question candidate to a generative model functionalized to output answer text. The acquired answer text is then provided to the user of the user terminal.