Geometric facial gender scoring: objectivity of perception

Area of Science:

• Computer Vision
• Psychology
• Biometrics

Background:

• Gender score, a continuum of perceived masculinity/femininity, is vital in psychology and cosmetic surgery.
• Facial structure and gender perception are linked to neurological disorders.
• Subjective gender scoring by humans is time-consuming and resource-intensive.

Purpose of the Study:

• Investigate geometric facial features used in human gender perception.
• Develop a mathematical model to objectively quantify facial gender perception.
• Mimic human cognitive judgments of facial masculinity and femininity.

Main Methods:

• Collected 3D face scans from 64 subjects using the 3dMDface scanner.
• Assigned gender scores to textureless 3D faces by 75 human raters.
• Analyzed geometric features, including Euclidean and geodesic distances between facial landmarks.

Main Results:

• Human gender scoring relies on a combination of distances between facial landmarks.
• Proposed mathematical model automatically assigns objective gender scores.
• Model achieved a high correlation (up to 0.895) with subjective human scores.

Conclusions:

• Objective, automated gender scoring of 3D faces is feasible using geometric analysis.
• The developed model accurately reflects human cognitive gender perception.
• This approach has potential applications in psychology, forensics, and medical fields.