A comprehensive AI-powered virtual wardrobe application that combines intelligent clothing classification, virtual try-on capabilities, personal wardrobe management, outfit advice, and 3D model, avatar generation. Built with a FastAPI backend and React frontend, featuring Gradio Space AI for realistic virtual try-ons, Roboflow for automatic clothing classification, OpenRouter LLM for outfit advice, and Pollinations AI for image and avatar generation, Trellis Gradio Space for 3D model generation.
🔗 Related Documentations: Frontend README | Backend README
📽️ Project Demo: View Smart Virtual Wardrobe Demo
We proudly presented this research work at, "2025 IEEE 7th International Conference on Computing, Communication and Automation (ICCCA)" held on November 28-30, 2025, at Galgotias University, Greater Noida, India through Hybrid Mode
Paper-ID with title: 1338 - Smart Virtual Wardrobe: AI-Powered Outfit Planner and Style Assistant
-
📄 View IEEE Conference Paper: 👉 Click to Open the Research Paper (PDF)
-
🏅 View Presentation Certificate: 👉 Click to Open Presentation Certificate (PDF)
-
📜 View Full Project Report: 👉 Click to Open Full Project Report (PHASE-1) (PDF)
-
🌐 IEEE Xplore Publication Link: 👉 Click to Open Paper's DOI Link
- 🤖 AI Virtual Try-On: Upload person and clothing images to generate photorealistic try-on results using the Gradio Virtual Try-On Space
- 👤 Profile Management: Users can update their profile and upload a profile photo stored on Cloudinary
- 📸 Smart Clothing Classification: Automatic categorization of clothing items using Roboflow AI
- 👕 Personal Wardrobe Management: Digitally organize, search, and manage your clothing collection
- 🧠 LLM Outfit Advisor: Upload clothing images and get AI-powered outfit suggestions via OpenRouter
- 🎨 Style Feed & Apparel Catalog: Browse/explore and save AI-generated style cards by filtering through an apparel product catalog (gender, season, color, article type, etc.)
- 🤖 Avatar Generator: Generate a 3D model/video (MP4 preview + interactive GLB viewer) from a clothing image using the TRELLIS Gradio space
- 🔐 Secure Authentication: JWT-based user authentication system
- ☁️ Cloud Storage: Images stored securely on Cloudinary CDN
- 📱 Responsive Design: Works seamlessly on desktop and mobile devices
- 🌓 Dark/Light Mode: Toggle between themes for comfortable viewing
Below are some screenshots showcasing the Smart-Virtual-Wardrobe core features:
- Frontend: React.js, Ant Design, React Router, React Context API, Axios, Three.js (GLB viewer)
- Backend: FastAPI, Python 3.12+, Uvicorn, Pydantic, JWT Authentication, Gradio Client
- AI Services: Gradio / Kolors Virtual Try-On, Roboflow Classification, OpenRouter LLM, Pollinations Image & Avatar Generation, TRELLIS 3D Generation
- Cloud Services: Cloudinary (Image Storage & CDN)
- Database: MongoDB Atlas (async via Motor)
- Python 3.12+
- Node.js 18+
- MongoDB Atlas account
- Cloudinary account
- Roboflow API key
- OpenRouter API key
- Pollinations API key (optional)
- HuggingFace token (optional, for TRELLIS 3D generation)
git clone https://github.com/Preveen369/Smart-Virtual-Wardrobe.git
cd Smart-Virtual-Wardrobecd backend
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# OR using Poetry
poetry install
poetry shell
# Set up environment variables
cp .env.example .env
# Edit .env with your API keysROBOFLOW_API_KEY=your_roboflow_api_key_here
CLOUDINARY_CLOUD_NAME=your_cloudinary_name
CLOUDINARY_API_KEY=your_cloudinary_key
CLOUDINARY_API_SECRET=your_cloudinary_secret
MONGODB_URI=your_mongodb_atlas_connection_string
JWT_SECRET=your_jwt_secret_key
# OpenRouter (LLM Outfit Advisor)
OPENROUTER_API_KEY=your_openrouter_api_key
OPENROUTER_BASE_URL=https://openrouter.ai/api/v1 # optional, default shown
OPENROUTER_MODEL=nvidia/nemotron-nano-12b-v2-vl:free # optional, default shown
OPENROUTER_FALLBACK_MODEL=gpt-4o-mini # optional, default shown
# Gradio Virtual Try-On space URL (optional, default shown)
GRADIO_TRYON_URL=https://ai-modelscope-kolors-virtual-try-on.ms.fun/
# TRELLIS multiple3D space URL (optional, default shown)
GRADIO_TRELLIS_URL=https://trellis-multiple3d.ms.fun/
# Pollinations image/avatar generation (optional)
POLLINATIONS_API_KEY=your_pollinations_api_key_here
# HuggingFace token for TRELLIS 3D space (optional)
HF_TOKEN=your_huggingface_token_hereStart the backend server:
uvicorn main:app --reload # port: 8000cd frontend
# Install dependencies
npm installStart the frontend development server:
npm run dev # port: 5173POST /register # User registration
POST /login # User login
GET /me # Get current user info
GET /profile # Get user profile
POST /profile # Create user profile
PUT /profile # Update user profile
POST /profile/photo # Upload profile photo
POST /api/try-on # Generate virtual try-on
GET /api/try-on/sessions # Get user's try-on history
POST /api/wardrobe/classify # Classify clothing image (Roboflow)
POST /api/wardrobe/items # Add new wardrobe item
GET /api/wardrobe/items # Get user's wardrobe items
GET /api/wardrobe/items/{id} # Get specific item
DELETE /api/wardrobe/items/{id} # Delete item
POST /api/outfit-advisor/analyze # Analyze outfit and get AI advice
POST /api/outfit-advisor/upload # Upload image for advisor
GET /api/outfit-advisor # List user's outfit advice history
GET /api/outfit-advisor/{id} # Get specific advice record
DELETE /api/outfit-advisor/{id} # Delete advice record
GET /api/favorites # List favorites (optional ?type= filter)
POST /api/favorites # Save a favorite
DELETE /api/favorites/{id} # Remove a favorite
GET /api/stylefeed # List style-feed cards (latest first)
GET /api/apparel/filters # Get available filter options
GET /api/apparel/products # Get filtered apparel products
The image-generation endpoint now lives under the style_feed router (see routers/style_feed.py), while avatars remain in routers/avatar.py.
GET /api/image/{prompt} # Generate image via Pollinations
POST /api/avatar # Generate 3D-style avatar via Pollinations
POST /generate-3d # Generate 3D model + GLB from image (TRELLIS)
Smart-Virtual-Wardrobe/
├── backend/
│ ├── main.py # FastAPI application entry point
│ ├── database.py # MongoDB connection (Motor async)
│ ├── cloudinary_config.py # Cloudinary folder configuration
│ ├── routers/
│ │ ├── auth.py # Authentication & profile endpoints
│ │ ├── avatar.py # Avatar generation endpoints
│ │ ├── tryon.py # Virtual try-on endpoints
│ │ ├── wardrobe.py # Wardrobe management endpoints
│ │ ├── outfit_advisor.py # LLM outfit advice endpoints
│ │ ├── favorites.py # Favorites endpoints
│ │ ├── style_feed.py # Style feed & image generation endpoints
│ │ ├── apparel.py # Apparel catalog endpoints
│ │ └── model3d.py # 3D model generation endpoints
│ ├── models/
│ │ ├── schemas.py # Pydantic models / schemas
│ │ └── database_ops.py # Database CRUD operations
│ ├── utils/
│ │ ├── base64_helpers.py # Image encoding utilities
│ │ └── apparel_only.csv # Apparel catalog dataset
│ ├── model_training_books/ # ML model training notebooks
│ ├── avatars_3D/ # Generated 3D assets (mp4 + glb)
│ ├── requirements.txt # Python dependencies
│ └── pyproject.toml # Poetry configuration
├── frontend/
│ ├── src/
│ │ ├── App.jsx # Main application component & routing
│ │ ├── context/
│ │ │ └── AuthContext.jsx # Authentication context provider
│ │ ├── components/
│ │ │ ├── Footer.jsx # Footer component
│ │ │ ├── GLBViewer.jsx # Interactive 3D GLB viewer (Three.js)
│ │ │ ├── ImageUpload.jsx # Image upload component
│ │ │ └── Logout.jsx # Logout button component
│ │ ├── pages/
│ │ │ ├── HomePage.jsx
│ │ │ ├── TryOnPage.jsx
│ │ │ ├── WardrobePage.jsx
│ │ │ ├── FavoritesPage.jsx
│ │ │ ├── HistoryPage.jsx
│ │ │ ├── ProfilePage.jsx
│ │ │ ├── OutfitAdvisorPage.jsx
│ │ │ ├── StyleFeed.jsx
│ │ │ ├── AvatarGeneratorPage.jsx
│ │ │ ├── HelpPage.jsx
│ │ │ ├── LoginPage.jsx
│ │ │ └── RegisterPage.jsx
│ │ ├── services/
│ │ │ └── api.js # Axios API service functions
│ │ └── assets/
│ │ └── ClothHangerIcon.jsx
│ ├── public/ # Public static files
│ ├── package.json # Node.js dependencies
│ └── vite.config.js # Vite configuration
├── screenshots/ # Application screenshots
├── reports/ # IEEE paper, certificate, project report
└── README.md # Project documentation
For detailed documentation on each major component, see the dedicated READMEs:
cd backend
pytest tests/ -vcd frontend
npm test- Register a new user account
- Upload clothing items to your wardrobe
- Use the classification feature to auto-categorize items
- Try the virtual try-on with person and clothing images
- Test search and filtering functionality
- Verify responsive design on mobile devices
# Frontend production build
cd frontend
npm run build
# Backend production
cd backend
uvicorn main:app --host 0.0.0.0 --port 8000- Development: Local MongoDB, Cloudinary dev environment
- Staging: MongoDB Atlas staging, Cloudinary staging
- Production: MongoDB Atlas production, Cloudinary production with CDN
- JWT token-based authentication
- Password hashing with bcrypt
- CORS configuration for frontend-backend communication
- Secure file upload validation
- Input sanitization and validation
We welcome contributions! Please follow these steps:
- Fork the repository
- Create a feature branch (
git checkout -b feature/AmazingFeature) - Commit your changes (
git commit -m 'Add AmazingFeature') - Push to the branch (
git push origin feature/AmazingFeature) - Open a Pull Request
- [✔️] Phase 1: Virtual try-on and Wardrobe Clothing Classification
- [✔️] Phase 2: Outfit recommendation and Weather-based outfit suggestions
This project is licensed under the MIT License - see the LICENSE file for details.
- Google for Gemini AI API
- Roboflow for clothing classification models
- Cloudinary for image storage and CDN
- MongoDB Atlas for database hosting
- FastAPI team for the excellent framework
- React community for UI libraries
For queries or suggestions:
- 📩 Email: [email protected]
- 🌐 LinkedIn: https://linkedin.com/in/preveen-s/
If you like this project, please consider giving it a ⭐ on GitHub!