Facial Asymmetry-Based Age Group Estimation: Role in Recognizing Age-Separated Face Images.

Area of Science:

• Computer Science
• Biometrics
• Artificial Intelligence

Background:

• Face recognition is challenged by aging variations due to complex changes in facial tissues.
• Existing methods often overlook age group estimation knowledge, limiting performance on age-separated faces.
• Facial asymmetry is an intrinsic, age-dependent feature relevant to both recognition and age estimation.

Purpose of the Study:

• To propose an age-assisted face recognition approach to mitigate challenges posed by aging variations.
• To leverage facial asymmetry for improved age group estimation and subsequent face recognition.
• To integrate age estimation insights into face recognition using deep convolutional neural networks (dCNNs).

Main Methods:

• Utilizing asymmetric facial dimensions to estimate the age group of a face image.
• Extracting deeply learned asymmetric facial features via a dCNN for face recognition.
• Integrating age group estimation knowledge into the dCNN for enhanced face recognition performance.

Main Results:

• The proposed age-assisted method significantly improves face recognition performance compared to algorithms without age integration.
• Experimental results on MORPH and FERET datasets demonstrate superior performance over existing state-of-the-art methods.
• The integration of age group estimation knowledge proved effective in handling aging variations.

Conclusions:

• The age-assisted face recognition approach effectively addresses aging variations by incorporating age group estimation.
• Facial asymmetry serves as a valuable feature for both age estimation and face recognition.
• The study highlights the benefit of multi-task learning or knowledge transfer in biometric systems.