Real-Time Gender Recognition for Juvenile and Adult Faces.

Area of Science:

• Computer Vision
• Biometrics
• Machine Learning

Background:

• Facial gender recognition has broad applications but struggles with unconstrained orientations and juvenile faces.
• Existing models face accuracy limitations due to variations in face angle and age-specific features.

Purpose of the Study:

• To develop a robust facial gender recognition system for juveniles, adults, and unconstrained-oriented faces.
• To address the limitations of current state-of-the-art models in real-world scenarios.

Main Methods:

• Utilized a progressive calibration network (PCN) for rotation-invariant face detection.
• Employed Gabor filters for edge and texture feature extraction, followed by meanDWT optimization to reduce redundancy and improve efficiency.
• Classified features using Naïve Bayes, Logistic Regression, and Support Vector Machines (SVM).

Main Results:

• The proposed meanDWT feature optimization method significantly improved system accuracy, reduced model size, and decreased computational time.
• The system demonstrated enhanced performance in recognizing gender across diverse facial orientations and age groups.
• Comparative analysis showed superior results against state-of-the-art techniques.

Conclusions:

• The developed facial gender recognition system offers improved accuracy and efficiency for real-world applications.
• The progressive calibration network and meanDWT optimization are effective in handling challenges posed by unconstrained faces and juvenile demographics.
• The study provides a comprehensive review of conventional and deep learning methods for facial gender recognition.