A Multifeature Learning and Fusion Network for Facial Age Estimation

Area of Science:

• Computer Vision
• Machine Learning
• Biometrics

Background:

• Facial age estimation is crucial for applications like video surveillance and social networking.
• Existing methods often overlook influential appearance features like gender and race, limiting accuracy.
• Integrating diverse features can enhance the robustness of age estimation models.

Purpose of the Study:

• To develop a compact, multi-feature learning and fusion method for accurate facial age estimation.
• To address the limitations of single-feature approaches in age estimation.
• To create a model efficient enough for deployment on resource-constrained devices.

Main Methods:

• Utilized three subnetworks to independently learn gender, race, and age features.
• Fused these complementary features to generate more robust representations.
• Employed a regression-ranking age-feature estimator to predict precise age values.

Main Results:

• The proposed method demonstrated superior effectiveness and efficiency on three benchmark facial age estimation databases.
• Achieved state-of-the-art performance compared to existing facial age estimation techniques.
• The model achieved a compact size with only a 20 MB memory overhead.

Conclusions:

• The multi-feature learning and fusion approach significantly enhances facial age estimation accuracy.
• The developed model is highly efficient and suitable for real-time applications on mobile and embedded systems.
• This method offers a practical solution for age estimation in diverse real-world scenarios.