MAGE-Vein: Multi-Instance Age and Gender Estimation from Finger Vein Images

Researchers introduced MAGE-Vein, a multi-task framework for estimating age and gender from finger vein images.

By combining information from three fingers and training age and gender jointly, it reduces noise and gender-related confounding.

On 402 subjects, the model achieved 6.12 years mean absolute error and a 0.880 correlation.

The findings suggest earlier poor results may have reflected dataset bias rather than limits of finger vein biometrics.