PantryMind – Smart AI for Food Waste Reduction

Turn your pantry into personalized meals — AI scans, detects, and generates recipes to cut food waste and save money.

Comment

🌱 Inspiration

Food waste is a growing global issue — according to the UN, nearly 1/3 of all food produced is wasted, leading to both financial loss and environmental harm. We were inspired to build PantryMind to empower households to use what they already have, cut down waste, and save money. The idea came from everyday situations: opening the fridge, finding ingredients close to expiry, and struggling to figure out what to cook.

🍽️ What it does

PantryMind is an AI-powered web app that helps households reduce food waste by:

  • 📷 Detecting pantry ingredients using YOLOv8 object detection.
  • 🧾 Extracting expiry dates with OCR to prioritize usage.
  • 🧠 Generating smart recipes using a RAG (Retrieval-Augmented Generation) system with FAISS + Sentence Transformers.
  • 📊 Tracking waste reduction and savings in a real-time analytics dashboard.

Simply put: You scan your pantry → PantryMind suggests personalized recipes → Your dashboard shows money, CO₂, and food saved.

⚙️ How we built it

  • Backend (FastAPI) – powered the AI pipeline and API routes.
  • YOLOv8 – for real-time ingredient detection.
  • EasyOCR – to scan expiry dates on packaging.
  • RAG System – built with all-MiniLM-L6-v2 + FAISS vector search on the RecipeNLG dataset (2.3M recipes).
  • Frontend (React + Vite + TailwindCSS) – for a clean, responsive UI with live dashboard updates.
  • Database (MongoDB) – to manage pantry items and analytics, with JSON fallback.
  • Optional GPT-3.5 Integration – to enhance and rewrite recipes creatively.

🚧 Challenges we ran into

  • Handling the huge RecipeNLG dataset (2.1GB) and converting it into a usable FAISS index.
  • Balancing speed vs accuracy for YOLOv8 detection on normal CPUs.
  • Ensuring fallback systems (JSON storage, RAG fallback if OpenAI unavailable).
  • Designing a modern UI/UX that is both functional and visually appealing within hackathon time limits.

🏆 Accomplishments that we're proud of

  • Successfully built a working end-to-end pipeline (scan → detect → generate recipes → update analytics).
  • Indexed millions of recipes for fast retrieval.
  • Designed a professional React frontend with real-time dashboard updates.
  • Built a scalable system that can run on both local machines and cloud deployments.

📖 What we learned

  • How to integrate multiple AI models (YOLO, OCR, RAG, GPT) into one seamless system.
  • Building and querying a FAISS vector database efficiently.
  • Optimizing AI inference for both speed and accuracy.
  • The importance of fallback systems in real-world applications.
  • Team collaboration under hackathon deadlines.

🚀 What's next for PantryMind – Smart AI for Food Waste Reduction

  • 📱 Mobile App Integration for quick pantry scanning on-the-go.
  • 🛒 Grocery store APIs to suggest buying only the missing items.
  • 🌍 Community-driven recipes where users can share and upvote waste-free ideas.
  • 🔔 Smart notifications when items are close to expiry.
  • ♻️ Expansion into restaurants and supermarkets to reduce commercial food waste.

PantryMind is not just a project — it’s a step toward smarter kitchens, sustainable living, and a greener planet. 🌍💚

Built With

+ 1 more
Share this project:

Updates