Character-Linked Image Recognition for Identifier Inference

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing communication and game applications lack an inbound-direction mechanism for attributing image information to character attributes, limiting the interaction between game characters and external images.

Innovation Solution

An electronic device and system that utilize image recognition models linked with character attributes to infer identifier information from image content, enabling the recognition and response to external images based on individual character traits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple image recognition models are stored for different character attributes, then the system can provide personalized image recognition for each character, but the storage requirements and system complexity increase

Engineering Contradiction:

Improvepersonalized image recognition capabilityVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent divides the image recognition task into multiple specialized models, each trained for a specific character attribute (e.g., character A's preferences, character B's preferences). This segmentation allows each model to be simpler and more specialized, improving recognition accuracy for specific attributes while maintaining manageable complexity through modular architecture.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent creates a universal framework that can handle multiple character attributes through a consistent interface and structure. The system provides multi-functionality by enabling the same basic recognition pipeline to work across different characters and attributes, reducing overall system complexity despite handling diverse recognition tasks.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Measurement precision

If image recognition models are trained with character-specific learning content, then the recognition accuracy for character attributes improves, but the data preparation and model training complexity increases

Engineering Contradiction:

Improveidentifier information inference accuracyVSAvoidmodel preparation ease

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent applies local quality by training each image recognition model with character-specific learning content tailored to that character's attributes and preferences. This localized training approach improves inference accuracy for each character while maintaining a standardized overall system architecture that simplifies the preparation process.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent performs preliminary action by pre-training image recognition models with character-specific learning content before deployment. This advance preparation allows the system to achieve high recognition accuracy when characters receive images, while the models are ready for immediate use without requiring complex real-time training.

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If the system processes images through multiple character-specific models, then the response personalization improves, but the processing time and computational load increase

Engineering Contradiction:

Improveresponse personalizationVSAvoidimage processing speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments the image processing task by routing images to specific character models based on the recognition task required. This segmentation enables parallel processing of different character attributions and reduces the computational burden on any single model, improving overall processing efficiency while maintaining personalized responses.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by selectively applying image recognition models only when needed for specific character interactions. The system processes images through character-specific models only when attribute recognition is required, rather than continuously processing all images through all models, thus reducing unnecessary computational load while maintaining response personalization.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12468962B2Electronic device, method, program, and system for identifier-information inference using image recognition model

Publication Date: 2025.11.11 CYGAMES INC

AI summary

Provided is an electronic device including: a storage unit for storing a plurality of image recognition models each of which is defined by an item of learning content unique thereto, with each of which at least one item of identifier information can be inferred by using the item of learning content, and each of which is linked with an item of attribute information unique thereto; a destination-attribute-information identifying unit that identifies at least one item of attribute information as a destination of image information from among items of attribute information stored in the storage unit, on the basis of an operation; an image-recognition-model selecting unit that selects the image recognition model linked with the identified item of attribute information; and an identifier-information inferring unit that inputs the image information to the image recognition model selected by the image-recognition-model selecting unit and that infers an item of identifier information.