AI-Driven Skincare Recommendation System Using Facial Analysis

Abstract

The increasing demand for personalized skincare solutions has encouraged the integration of Artificial Intelligence (AI), Machine Learning (ML), and Computer Vision technologies in the skincare industry. Traditional skincare selection methods often rely on generalized advice, trial-and-error approaches, or manual consultation, which may not provide accurate recommendations for different skin conditions. To address this challenge, this paper presents Beauty-Skin AI, an AI-driven skincare recommendation system that analyses facial images and generates personalized skincare insights, routines, and product suggestions. The proposed system combines facial image processing, heuristic feature extraction, Convolutional Neural Networks (CNNs), dataset profile classification, confidence scoring, and rule-based recommendation techniques to analyze skin characteristics such as oiliness, pigmentation, redness, texture irregularities, and dark circles. The system supports multiple image acquisition methods, including image upload, live camera capture, image URLs, and dataset image paths, ensuring flexibility and accessibility. Based on extracted features and predicted skin type, the recommendation engine generates personalized AM/PM skincare routines and suitable skincare products using dataset-driven filtering and lightweight machine learning techniques. Additionally, IndexedDB-based history management enables users to store and compare previous analysis records for long-term tracking. The proposed system demonstrates how hybrid AI techniques can improve personalization, explainability, and usability in intelligent skincare recommendation systems while maintaining scalability for future enhancements and practical real-world deployment.