A cutting-edge, Gen AI-powered virtual try-on web application that enables users to upload a person (model) image and a clothing item image to generate photorealistic try-on previews in seconds. Built using Google Gemini (Generative AI) and FastAPI, this tool delivers high-fidelity outfit visualization, facial identity preservation, background replacement, and seamless garment rendering—perfect for fashion tech, e-commerce, and virtual fitting room experiences.
- Upload any person image and clothing item
- Generate AI-based try-on images with high fidelity
- Preserves facial identity and garment texture
- Automatically removes and replaces the background
- Responsive UI with dark/light mode
- View and save previous try-on history in session only
| 🖥️ Frontend | ⚙️ Backend | 🤖 AI & Processing |
|---|---|---|
| React.js | FastAPI | Google Gemini API (Generative AI) |
| Ant Design (UI Library) | Uvicorn (ASGI) | Image-to-Image Inference |
| Axios (API Calls) | Python 3.12+ | Base64 Encoding/Decoding |
| React Toastify | Pydantic | Multimodal Content Handling |
git clone https://github.com/narender-rk10/Gen-AI-Virtual-Try-On-Clothes.git
cd Gen-AI-Virtual-Try-On-Clothescd backend
poetry install
poetry shellCreate a .env file and add:
GEMINI_API_KEY=your_gemini_api_key_here
Run the server:
uvicorn main:app --reloadcd frontend
npm install
npm run devPOST /api/try-on
person_image: UploadFile (image)cloth_image: UploadFile (image)instructions,model_type,gender,style,garment_type: FormData
/frontend # React + Ant Design UI
/backend # FastAPI + Gemini AI API
Pull requests are welcome. For major changes, please open an issue first to discuss what you’d like to change.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
This was built to explore fashion AI possibilities using state-of-the-art multimodal LLMs like Google Gemini.