Facial Recognition Threshold Adjustment for Distinctive Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Facial recognition systems face issues with false positive matches due to the use of a uniform threshold similarity score, which can lead to unauthorized access for users with less distinctive facial features, as they may generate higher similarity scores compared to those with more distinctive features.
Innovation Solution
The system adjusts the threshold similarity score based on the distinctiveness of each user's facial features by comparing them against a diverse population, setting a higher score for less distinctive features and a lower score for more distinctive features to maintain a consistent false positive match rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a uniform threshold similarity score is used for all users, then the system is simple to operate and implement, but users with less distinctive facial features experience false positive matches and unauthorized access
Solution Approach 1:
The patent applies local quality by customizing the threshold similarity score for each individual user based on their specific facial feature distinctiveness. Instead of using a uniform threshold for all users, the system determines a personalized threshold for each user by comparing their facial features against a diverse population, ensuring that each user's authentication threshold reflects their unique facial characteristics.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting the threshold similarity score parameter based on the distinctiveness metric. The system calculates a distinctiveness score for each user by comparing their facial features to a diverse population, then uses this distinctiveness score to determine an appropriate threshold similarity score, transforming the fixed parameter into a variable one that adapts to individual user characteristics.
2Reliability
If the threshold similarity score is increased to reduce false positives, then authentication accuracy improves, but legitimate users with distinctive features may be incorrectly rejected
Solution Approach 1:
The system applies local quality by tailoring the threshold similarity score to each user's specific facial distinctiveness. Users with more distinctive facial features receive lower thresholds, making access easier for them, while users with less distinctive features receive higher thresholds, reducing their false positive rate. This localized approach ensures that each user experiences optimized authentication conditions based on their individual characteristics.
Solution Approach 2:
The patent employs feedback by using the distinctiveness score as a feedback mechanism to adjust the threshold similarity score. The system first measures the distinctiveness of each user's facial features by comparing them against a diverse population, then uses this measurement as feedback to set an appropriate threshold that balances security and accessibility for that specific user.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In general, aspects of the present disclosure are directed to techniques for adjusting the threshold similarity score to match the facial features of an authorized user in a facial recognition system. Instead of a uniform threshold similarity score, representations of authorized faces enrolled in the computing device may be assigned a custom threshold similarity score based at least in part on the distinctiveness of the facial features in the representations of the authorized faces. A computing device can determine a distinctiveness of facial features of an enrolling user, and can determine a threshold similarity score associated with the facial features of the enrolling user based at least in part on the distinctiveness of the facial features of the enrolling user.