Implementasi Metode WP dalam Pemilihan Skincare Menggunakan Computer Vision Berbasis Mobile

Abstract

Selecting the right skincare products tailored to one's skin condition is challenging due to the vast array of available products and users' limited understanding of their own skin characteristics. Improper selection can lead to issues such as irritation, acne, and damage to the skin barrier. This study developed the "SkinPick" mobile application as a decision support system to assist users in obtaining skincare recommendations based on their facial skin type and condition. The Weighted Product (WP) method was employed to generate recommendations by considering six criteria: normal, dry, oily, sensitive, combination, and acne-prone skin. Additionally, the application utilizes Roboflow-based Computer Vision technology to identify skin conditions through facial scanning. SkinPick was developed using Flutter, Dart, and Firebase and subsequently tested using the Black Box Testing method. Test results confirmed that all application functions operated as intended. WP calculations identified Niacinamide as the optimal alternative with a score of 0.1899, followed by Hyaluronic Acid at 0.1880. Expert validation yielded a 93.33% agreement rate. The Computer Vision model achieved an overall accuracy of 90.1%, mAP of 89.2%, precision of 91.5%, recall of 87.4%, and an F1-score of 89.4%. These results demonstrate that SkinPick is capable of providing skincare recommendations in a more objective and targeted manner.