GlamSync
GlamSync What if your laptop camera could do more than just show you your reflection? We built GlamSync AI , a personal beauty styling coach that combines Gemma 4, OpenCV, and real-time AR to help users understand what makeup and styling choices actually suit them. The Problem Choosing makeup that works with your individual face, skin undertone, and outfit can be surprisingly difficult. Most…
GlamSync AI is a personal beauty styling coach that uses advanced technology to help users find makeup and styling choices that suit their unique facial features. The AI system combines Gemma 4, OpenCV, and real-time AR to analyze an individual's outfit, facial geometry, and skin tone.
First, the Outfit Color Analysis feature uses OpenCV and K-Means clustering to extract dominant colors from a user's outfit and understand its color palette. Next, Facial Analysis employs OpenCV YuNet to examine facial geometry, and CIELAB-based ITA colorimetry and skin segmentation estimate skin tone and undertone.
Gemma 4 then takes all of this information into account, along with the user's occasion, budget, and available products, to generate personalized styling looks and step-by-step recommendations. The Real-Time AR Coach takes things further by tracking facial landmarks using MediaPipe's 468-point Face Mesh, and overlaying guidance for areas such as cheeks, eyelids, eyeliner, and lips. The user can follow the instructions while applying their makeup, receiving voice guidance for a hands-free experience.
The developers behind GlamSync AI wanted to move beyond traditional beauty applications that offer static filters or generic recommendations. Instead, they aimed to create an interactive AI styling coach that could truly understand a user's facial features and provide hands-on guidance for makeup application. The system is built using React, TypeScript, Vite, FastAPI, Python, Gemma 4, OpenCV, MediaPipe, NumPy, and more, and is available for free on GitHub.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.