Personalized Facial Wrinkle Distribution Analysis Using Backpropagation Neural Network (BPNN)

Abstract

Facial wrinkle distribution is an important indicator of aging and lifestyle. This study proposes a personalized wrinkle classification system using a Backpropagation Neural Network (BPNN) based on segmented facial areas such as the forehead, eyes, cheeks, and mouth. After preprocessing and feature extraction, the BPNN model is trained to classify wrinkle severity into two categories: high and medium. Evaluation results show that the model performs well, particularly in detecting the Medium class, achieving a precision of 0.8438 and a recall of 0.9310, while for the High class, the precision is 0.8333 and the recall is 0.6667. These findings indicate that the BPNN architecture is effective and reliable in facial wrinkle classification, with potential applications in dermatology, cosmetic analysis, and digital forensics.