Multi-Model Facial Recognition Threshold Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition technologies require inefficient manual configuration of decision-making thresholds for multiple models, leading to lower accuracy in adapting to varying scenarios and risk levels.
Innovation Solution
A method that uses joint testing of auxiliary detection models and feature-matching models to dynamically determine decision-making thresholds, improving adaptability and accuracy by considering the mutual influence between different algorithm models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual configuration of decision-making thresholds is used for each model, then the system can be operated, but the efficiency is low and accuracy is reduced
Solution Approach 1:
The system automatically determines decision-making thresholds through joint testing of multiple models without manual intervention. The threshold determination module performs self-service by autonomously analyzing test results and generating optimal threshold values, eliminating the inefficiency of manual configuration while maintaining high accuracy through systematic multi-model evaluation
Solution Approach 2:
The system performs preliminary joint testing with sample images before actual facial recognition operations. By pre-determining optimal thresholds through advance testing and analysis, the system prepares the best configuration in advance, avoiding the need for manual threshold setting during operation and ensuring high accuracy from the start
2Reliability
If multiple models are used for facial recognition, then comprehensive recognition results are obtained, but the complexity of threshold configuration increases
Solution Approach 1:
The system merges the threshold determination process across multiple models by performing joint testing simultaneously. Instead of configuring thresholds for each model separately, the threshold determination module combines all models in a unified testing framework, reducing configuration complexity while maintaining comprehensive recognition through coordinated multi-model evaluation
Solution Approach 2:
The threshold determination module serves multiple functions: it tests multiple models, analyzes their interactions, determines optimal thresholds for all models, and validates the configuration. This universal module handles the complex multi-model threshold configuration task through a single integrated process, simplifying the overall system complexity while ensuring reliable comprehensive recognition
Data Source
AI summary
Embodiments of the disclosure provide a recognition method, apparatus, device, and storage medium and relates to the field of artificial intelligence technology. The method includes obtaining a second decision-making threshold of a feature-matching model in a target scenario by joint testing of the feature-matching model and auxiliary detection model. The method takes into full consideration the mutual influence between different algorithm models in a scenario when multiple algorithm models are used for facial recognition. Compared to manually setting a decision-making threshold for each algorithm model independently, the methods in the disclosure are more adaptable to changing scenarios and scenarios with multiple models used in facial recognition. This improves the accuracy and efficiency of the obtained decision-making thresholds, thereby enhancing the accuracy of multi-model facial recognition.