Adaptive Enrollment Database for Face Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face recognition in wearable and mobile devices experiences decreased accuracy due to changes in user appearance over time, such as makeup style, hair style, or facial hair, leading to automated authentication challenges.
Innovation Solution
A processor-implemented method with an adaptively updated enrollment database that extracts input feature vectors, determines their inclusion in a changeable enrollment range based on threshold distances from enrolled feature vectors, and selectively enrolls or updates these vectors to maintain accurate authentication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated authentication is performed using a fixed enrollment database, then authentication speed is maintained, but authentication accuracy decreases due to changes in user appearance over time
Solution Approach 1:
The enrollment database is transformed from a static structure to a dynamic one that automatically updates with new feature vectors. The system continuously adapts the enrollment database by adding new feature vectors when they fall within the changeable enrollment range, enabling the system to track and accommodate appearance changes over time while maintaining authentication accuracy.
Solution Approach 2:
The system implements a feedback mechanism where authentication results and new feature vectors are used to update the enrollment database. When a new feature vector is obtained through authentication, the system determines whether to enroll it based on distance calculations, and the enrollment database is subsequently updated, creating a closed-loop system that continuously improves accuracy based on accumulated data.
2Measurement precision
If the enrollment database is updated frequently to reflect appearance changes, then authentication accuracy is improved, but system complexity increases
Solution Approach 1:
The system manages complexity by changing key parameters such as the threshold distance (maximum enrollment distance) and minimum enrollment distance to control when and how feature vectors are enrolled. These parameter adjustments allow the system to adapt to appearance changes while maintaining manageable complexity through configurable thresholds rather than complex algorithms.
Solution Approach 2:
The enrollment database performs self-service by automatically determining whether to enroll new feature vectors based on distance calculations and predefined criteria. The system autonomously updates itself without requiring manual intervention or complex external management, reducing overall system complexity while maintaining high authentication accuracy.
3Reliability
If a strict threshold is used for enrollment to prevent false acceptance, then security is maintained, but false rejection rate increases due to appearance variations
Solution Approach 1:
The system uses partial enrollment actions by selectively adding only those feature vectors that meet specific criteria (falling within the changeable enrollment range). This partial action approach allows the system to maintain security by being selective about what is enrolled, while simultaneously improving acceptance rates by accumulating enough variation to cover appearance changes over time.
Data Source
AI summary
A method and apparatus with an adaptively updated enrollment database (DB) are provided. A method with an adaptively updated enrollment database (DB) includes extracting an input feature vector from an input image, determining whether the input feature vector is included in a changeable enrollment range, with the changeable enrollment range being determined based on a threshold distance from each of plural enrolled feature vectors in the enrollment DB, and with the enrolled feature vectors corresponding to enrolled images, determining whether to enroll the input feature vector in the enrollment DB in response to the input feature vector being determined as being included in the changeable enrollment range, and in response to a result of the determining of whether to enroll the input feature vector being to enroll the input feature vector, selectively enrolling the input feature vector in the enrollment DB.