Face Template Balancing for Facial Recognition Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional facial recognition algorithms often result in false positives when matching images of individuals with similar features, leading to incorrect tagging and reduced accuracy in social network systems.

Innovation Solution

A method and system that generate face templates for images, match them to users, and flag mismatched images as negative templates, using user feedback to differentiate between correctly and incorrectly matched images, thereby improving the accuracy of facial recognition by balancing positive and negative templates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional facial recognition algorithms are used to match images to users, then the matching process can be performed quickly and automatically, but false positives occur when matching images of individuals with similar features, leading to reduced accuracy

Engineering Contradiction:

Improvematching speedVSAvoidmatching accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent segments the face template into multiple regions (e.g., eyes, nose, mouth, cheeks) and analyzes differences in each region separately. This segmentation allows the system to identify subtle distinguishing features that conventional algorithms miss, thereby improving matching accuracy while maintaining automated processing efficiency.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality analysis by examining specific facial regions with different levels of detail. Certain regions are analyzed more closely than others based on their discriminative power. This approach enables the system to achieve higher accuracy by focusing computational resources on the most informative local features rather than treating the entire face uniformly.

Inventive Principle:

Principle #3Local quality

2Device complexity

If only positive face templates are used for matching, then the system can operate with simpler logic, but it cannot effectively distinguish between similar individuals, leading to false positives

Engineering Contradiction:

Improvesystem complexityVSAvoiddistinguishing accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces asymmetry by treating positive and negative face templates differently in the matching process. Negative templates (from mismatched images) are used to identify and exclude false positives, creating an asymmetric evaluation process that significantly improves distinguishing accuracy between similar individuals while adding manageable complexity.

Inventive Principle:

Principle #4Asymmetry

Solution Approach 2:

The patent implements feedback mechanisms where mismatched images are fed back into the system to generate negative templates. This feedback loop continuously refines the matching algorithm by learning from errors, enabling the system to progressively improve its ability to distinguish between similar individuals without requiring complete system redesign.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS10229311B2Face template balancing

Publication Date: 2019.03.12 GOOGLE LLC

AI summary

Implementations generally relate to face template balancing. In some implementations, a method includes generating face templates corresponding to respective images. The method also includes matching the images to a user based on the face templates. The method also includes receiving a determination that one or more matched images are mismatched images. The method also includes flagging one or more face templates corresponding to the one or more mismatched images as negative face templates.