VirtuFit: AI-Based Virtual Try-On Using Avatar and Face Mapping

Online fashion shopping lacks the tactile experience of trying clothes before purchase, causing uncertainty in fit and style.VirtuFit proposes a lightweight, AI-based virtual try-on (VTON) system that allows users to visualize garments on themselves using only a webcam.The system detects body landmarks using MediaPipe, estimates measurements, generates a 2D avatar through Canvas, and maps the user's face using OpenCV.Garments are automatically selected based on predicted body proportions and applied to the avatar in real time.Developed using React.js,Flask, and MongoDB, VirtuFit delivers an inter-active, realistic virtual try-on experience with an average latency of 2.1 seconds and 89% size accuracy.This paper discusses its architecture, methodology, performance, and future potential for accessible virtual fitting solutions.