In-Store Video Content Generation for Personalized Purchase Nudges

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods fail to effectively increase customer willingness to purchase by simply displaying commodity information, necessitating a more personalized approach to present relevant information.

Innovation Solution

A system utilizing cameras and display devices in a store to capture and analyze customer interactions with commodities, employing machine learning models to generate and retrain content based on transitions in customer actions, thereby identifying effective information to present.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If commodity information is displayed on accounting machines or similar devices, then customers can receive information about commodities, but customer willingness to purchase is not effectively increased

Engineering Contradiction:

Improvecommodity information deliveryVSAvoidcustomer purchase willingness

Core Design Contradiction:

Loss of informationVSProductivity

Solution Approach 1:

The system continuously monitors customer actions through video capture and analyzes transitions from pre-content to post-content states. This feedback loop enables the machine learning model to learn which content effectively influences customer behavior, allowing iterative improvement of information presentation strategies to increase purchase willingness

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system dynamically adjusts content parameters based on analyzed customer actions and transitions. By modifying content characteristics (such as product details, promotions, or information framing) based on learned patterns from video analysis, the system optimizes information delivery to maximize customer purchase willingness

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If a machine learning model is used to generate content based on customer actions, then personalized information can be presented, but the system complexity increases

Engineering Contradiction:

Improvepersonalized content generationVSAvoidsystem structure

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The machine learning model automatically analyzes customer actions from video data and generates appropriate content without requiring manual programming of response rules. The system self-adjusts and improves its content generation capabilities through continuous learning from observed customer behavior transitions, reducing the need for complex manual configuration

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system replaces complex manual decision-making mechanisms with automated machine learning models that process video data and generate content automatically. This substitution of mechanical/rules-based systems with intelligent algorithms simplifies the overall system architecture while enabling sophisticated personalized content generation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If video capture and analysis are performed continuously to analyze customer actions, then accurate identification of customer interests is achieved, but energy consumption and processing time increase

Engineering Contradiction:

Improvecustomer action identification accuracyVSAvoidvideo processing energy consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary content generation and display before final purchase decisions are made. By analyzing customer actions in the time window between content display and potential purchase, the system identifies interest transitions and adjusts content strategies proactively, improving measurement precision without requiring continuous processing throughout the entire customer journey

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system analyzes video data selectively by focusing on specific time windows around content display and identifying significant action transitions rather than processing every frame continuously. This partial action approach maintains adequate measurement precision for identifying customer interests while significantly reducing overall energy consumption and processing requirements

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12482297B2Computer-readable recording medium storing information processing program, information processing method, and information processing apparatus

Publication Date: 2025.11.25 FUJITSU LTD

AI summary

A non-transitory computer-readable recording medium storing an information processing program for causing a computer to execute a process including: acquiring a video obtained by capturing an inside of a store; identifying a first action of a person on a commodity disposed within the store by analyzing the acquired video; generating a content by inputting the video to a machine learning model; outputting the generated content to a terminal disposed within the store; identifying a second action of the person on the commodity by analyzing a video after the content is output to the terminal; and retraining the machine learning model, based on a transition from the first action to the second action, the generated content, and an image of the commodity as a target of the identified second action.