MenuMind is an intelligent food recognition and recommendation web application that identifies dishes from images using a CNN multiclass classifier and provides smart recommendations — blogs, videos, and related dishes — powered by Google APIs.
MenuMind bridges computer vision and web intelligence. Upload or capture an image of any dish, and the system:
- Analyzes the image using a trained CNN multiclass classifier to predict the dish name.
- Fetches related content — including blog posts, recipe ideas, and videos — via Google APIs.
- Displays results in an elegant UI with interactive sections for learning and discovery.
- React.js — Dynamic UI and smooth interactivity
- Tailwind CSS — Clean, responsive, and modern styling
- Axios — API communication between frontend and backend
- FastAPI (Python) — Handles ML model inference and image uploads
- Express.js (Node.js) — Manages recommendation API and external data fetching
- Google APIs — For blog, video, and content recommendations
- CNN Multiclass Classifier — Built with PyTorch & TorchVision to identify dishes
- EfficientNet-B0 base model for fast and accurate classification
✅ Real-time dish recognition from images
✅ Intelligent blog and video recommendations
✅ Dual backend architecture (FastAPI + Express)
✅ Modern UI using Tailwind and React
✅ Deployed on Render and Vercel for scalability
- Upload Image → Frontend sends the file to the FastAPI endpoint.
- ML Model Predicts → CNN model classifies the dish.
- Fetch Recommendations → Express backend fetches related blogs/videos.
- Render Results → React UI displays the predicted dish and recommended content beautifully.