Retail Information Processing for Skeleton-Based Interest Detection Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating detection rules for each product to detect customers highly effective in customer service is time-consuming and labor-intensive, making it difficult to identify such customers due to the vast number of products, as actions representing interest vary by product type.
Innovation Solution
A system that generates detection rules based on past actions, presence or absence of product purchase, and basic movements for multiple stages of interest, without relying on specific product types, using cameras and information processing devices to analyze skeletal information and generate rules correlating customer actions with product interest.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detection rules are generated manually for each product to detect customers highly effective in customer service, then the detection accuracy and relevance can be improved, but the time consumption and labor intensity increase significantly
Solution Approach 1:
The system enables automatic generation of detection rules by having the computer analyze image data, identify skeletal joint positions, recognize actions, and generate rules autonomously without manual intervention. This self-service approach eliminates the time-consuming manual rule creation process while maintaining detection accuracy through automated image analysis and action recognition algorithms.
Solution Approach 2:
The patent replaces the mechanical manual process of creating detection rules with an automated information processing system. The computer automatically processes image data, extracts skeletal information, identifies actions, and generates detection rules, substituting the manual mechanical process with automated computational methods that are both faster and scalable.
2Reliability
If detection rules are created for each product considering varying actions, then the specificity and effectiveness of customer detection can be improved, but the device complexity and operational difficulty increase
Solution Approach 1:
The system employs a universal action recognition framework that can detect multiple types of customer actions (looking, picking up, touching, etc.) using a single integrated processing pipeline. The skeleton-based action recognition model serves multiple product categories and action types without requiring separate specialized systems for each, thereby reducing overall system complexity while maintaining detection effectiveness.
Solution Approach 2:
The patent uses parameter changes in the form of configurable action thresholds and skeletal joint position criteria to adapt the same detection system to different product types and customer actions. By adjusting parameters such as action duration thresholds, spatial proximity thresholds, and skeletal pose criteria, the system can effectively detect various actions without increasing structural complexity.
3Measurement precision
If manual rule creation is performed for each product, then the detection precision for specific product types can be improved, but the productivity and efficiency of the detection system decrease
Solution Approach 1:
The system performs preliminary action by automatically analyzing a large corpus of image data and pre-generating detection rules before actual customer service operations. The automated rule generation process prepares detection rules in advance based on learned patterns from historical data, enabling the system to operate efficiently during actual use without requiring manual rule creation at the time of detection.
Solution Approach 2:
The patent substitutes the manual mechanical process of rule creation with automated computational methods that process image data, extract features, and generate rules algorithmically. This substitution dramatically increases productivity by eliminating human labor from the rule generation process while maintaining or improving detection precision through sophisticated image analysis algorithms.
Data Source
AI summary
An information processing device obtains each piece of image data captured within a period of time from entering until exiting of a person at a store. The information processing device identifies joint positions of a skeleton related to the person by analyzing each piece of the image data. The information processing device identifies, as an action which indicates a degree of interest of the person in the product, an action performed by the person to a product in the store from the entering until the exiting, on a basis of the joint positions of the skeleton. The information processing device generates a detection rule that correlates the identified action and the product with each other.