Facial Age and Gender Prediction using Deep Learning

In society, gender and age have two face characteristics that are highly significant. The main objectives of automated categorization are to create intelligent systems that can effectively learn and identifying faces. Deep learning methods are frequently employed in most studies and are also utilized to enhance performance. These applications use biometrics, which is generally utilized for security, as a crucial component. Emphasizing the significance of this technology, the paper discusses its potential to enhance everyday life through applications such as intelligence agencies, CCTV cameras, policing, and matrimony sites. One basic biometric technique that has gained prominence in the field of study is the face modality. The work's primary objective is intended to advance a system that recognizes the gender and age of a facial image using CNNs. We explore efficiency of custom CNN architectures versus utilizing established CNN architectures as feature extractors. Our strategy makes use of the UTK Faces dataset for gender prediction and face Age datasets for age estimation. From previous works the gender and age accuracy were 86% and 75% respectively. We obtain 89.0% accurate for gender and 78.0% accurate for age estimates by utilizing deep learning techniques. The results obtained indicate the usefulness of several pre-trained models for the task and show the value of deep learning for estimating gender and age in faces.