Renzoe Match: A Data-driven Beauty Product Matching System

This project aimed at implementing a data-driven method to match consumer skin tone with beauty products such as foundations or concealers. Key technology includes computer vision, regression machine learning model (XGBoost), and Bayesian Optimization. This skin tone matching system extracts consumers' skin tone data from their selfie and processes them to adjust color distortion and analyze skin tone profiles to recommend 3 best matching foundations. Due to the pandemic, consumers have become more reluctant to visit cosmetic stores in person. One's skin tone can be changed due to extensive outdoor activities, aging, or cosmetic treatments. This method is expected to provide a hassle-free and accurate replacement for in-person foundation matching.