Face Spoofing, Age, Gender and Facial Expression Recognition Using Advance Neural Network Architecture-Based...

Area of Science:

• Computer Science
• Biometrics
• Healthcare Technology

Background:

• Increasing demand for soft-biometric devices in daily life.
• Emergence of face biometrics in healthcare for managing patient and staff data.
• Need for secure digital systems to reduce paperwork and improve access to medical information.

Purpose of the Study:

• To propose a novel soft-biometric methodology for a secure biometric system in healthcare.
• To enhance the security and efficiency of managing sensitive medical information.
• To develop a system capable of classifying facial attributes like age, gender, expression, and detecting spoofing.

Main Methods:

• Utilized a 5-layer U-Net-based architecture for accurate face detection.
• Employed Alex-Net-based architecture for the classification of facial information (age, gender, expression, spoofing).
• Evaluated the proposed methodology on six benchmark datasets: NUAA, CASIA, Adience, IOG, CK+, and JAFFE.

Main Results:

• Achieved high accuracy rates: 94.17% for spoofing detection, 83.26% for age classification, 95.31% for gender classification, and 96.9% for facial expression classification.
• The proposed model demonstrated superior performance compared to existing state-of-the-art methodologies.
• Modifications in the proposed model led to significant improvements in overall results.

Conclusions:

• The developed soft-biometric system offers a secure and effective solution for healthcare applications.
• The methodology shows strong potential for future soft-biometric based applications, particularly in sensitive environments.
• The system's high accuracy in detecting various facial attributes and spoofing ensures robust data security and management.