Virtual Try-On and Size Suggestion Capstone Project

Introduction

This capstone project is a comprehensive AI-powered fashion technology solution that addresses two critical challenges in online shopping: uncertainty about garment fit and difficulty visualizing how clothes will look. By combining advanced computer vision and machine learning techniques, our system provides users with personalized size recommendations and realistic virtual try-on experiences.

The platform consists of two main components:

• Frontend: A modern, responsive web application built with React and Vite
• Backend API: A production-ready FastAPI service that handles ML model inference and image processing

Problem Statement

Online shopping faces significant obstacles:

• High return rates due to incorrect sizing
• Inability to physically try on clothes before purchase
• Lack of personalized recommendations based on body measurements
• Inconsistent sizing standards across brands

Our Solution

We developed an integrated system that:

1. Predicts optimal clothing sizes using machine learning models trained on body measurements (age, height, weight)
2. Generates realistic virtual try-on images using state-of-the-art diffusion models (OOTDiffusion) and Google Vertex API
3. Delivers a seamless user experience through an intuitive web interface

Acknowledgement

@article{xu2024ootdiffusion,
  title={OOTDiffusion: Outfitting Fusion based Latent Diffusion for Controllable Virtual Try-on},
  author={Xu, Yuhao and Gu, Tao and Chen, Weifeng and Chen, Chengcai},
  journal={arXiv preprint arXiv:2403.01779},
  year={2024}
}

Under Adivsory

• Ms. Sam Sreyleak

Teammember

• Both Chealean
• Chheang Sovanpanha
• Huy Visa
• Ly Kimkheng
• Nem Vuthsovannath