Vyom Try-On: AI-Powered Virtual Fitting Room

Vyom Try-On is an AI-based virtual fitting room implemented as a client-side single-page application developed in React 19 and TypeScript. Based on generative try-on technology, the system delivers interactive garment try-on by providing a seamless integration of inputs from a webcam or photo upload, processing of the image with HTML5 Canvas, and connection to the Gemini-2.5-flash-image model through a secure proxy layer. The architectural design of the application is centered around a user-friendly e-commerce platform consisting of a product catalog, shopping cart, user signup/login, and an administrator dashboard. The application utilizes Context API for state management and a localStorage-based Object-Relationship mapping library for data persistence without a dedicated backend database. In this paper we describe the end-to-end development and deployment of the application including system design, implementation details, empirical assessment with quantitative metrics, user experience evaluation, and detailed reproducibility documentation. Reporting of methodology, simulated experimental results, user-centric evaluation, and open-source resources are provided in the paper with an aim to disseminate practical knowledge about virtual fitting room systems and reproducible applied machine learning for online retail.