MEMPREDIKSI USIA DAN JENIS KELAMIN MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORKS

Abstract

The main objective of this study was to develop a method for estimating the
age and gender of a person based on facial images, using CNN in-depth
training that can accurately recognize age and gender. Information that is
extracted can be useful in, for example, security or commercial applications.
This is a difficult estimation problem, because the only information we have is
a picture, that is the look of that person.
The next aspect of this study I focused on incorporating architecture for age
and gender recognition to take advantage of gender-specific age
characteristics and age-specific gender characteristics inherent in the image.
This comes from the observation that sex classifications are tasks that are
inherently easier than age classifications, because both fewer and fewer
potential classes and more prominent intra-gender facial variations. With the
training of different age classifiers for each gender I found that I could improve
the performance of age classifications, even though gender classification did
not see significant results.