Dynamic Classifier Selection for Face Authentication Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face authentication systems face challenges in achieving high accuracy across various circumstances due to a trade-off between generalization and identification performance, and the lack of a condition for selecting classifiers, leading to degraded authentication accuracy when user conditions change or are not accounted for.
Innovation Solution
An authentication apparatus that extracts face features, derives specific conditions based on extracted features and user information, and selects appropriate classifiers to authenticate users, allowing for dynamic adaptation to changing user conditions and environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a classifier is designed to achieve high generalization performance for various face authentication scenarios, then the classifier can handle diverse authentication conditions, but the identification performance under specific conditions deteriorates
Solution Approach 1:
The patent implements dynamic classifier selection where the system switches between multiple classifiers based on detected authentication conditions. The controller selects different classifiers from a plurality of available classifiers depending on factors such as lighting conditions, face orientation, or other environmental factors, allowing the system to adapt dynamically to varying conditions while maintaining high identification accuracy for each specific scenario
Solution Approach 2:
The system changes the parameter of classifier selection based on authentication conditions. By detecting conditions such as lighting, angle, or other factors and using these as selection criteria, the system adjusts which classifier is active, thereby optimizing identification performance for each specific condition while maintaining overall versatility
2Measurement precision
If multiple classifiers specialized in particular conditions are used, then authentication accuracy under specific conditions improves, but the condition for selecting classifiers is not always given before authentication
Solution Approach 1:
The system performs preliminary detection of authentication conditions before actual authentication occurs. The controller detects conditions such as lighting, face orientation, or environmental factors in advance, and uses this preliminary information to select the appropriate classifier before the authentication process begins, ensuring optimal accuracy is achieved
Solution Approach 2:
The authentication system performs self-selection of classifiers based on automatically detected conditions. The controller monitors authentication conditions and autonomously determines which classifier to use without requiring external intervention or pre-programmed condition specifications, making the system adaptive and operationally simple
Data Source
AI summary
An authentication apparatus comprises an extraction part configured to extract a feature of a face from a face image of a user; a plurality of classifiers configured to identify the user by determining whether or not the feature extracted by the extraction part matches a feature of a face stored in a storage apparatus; a deriving part configured to derive a specific condition based on at least one of the feature extracted by the extraction part and information for authenticating the user; a selecting part configured to select at least one of the classifiers based on the specific condition; and an authentication part configured to authenticate the user when the selected classifier determines that the feature extracted by the extraction part matches the feature of the face stored in the storage.