nudge.

Less Waste, More Taste. By showing users what's in their fridge right now, nudge helps prevent food from being forgotten and thrown away.

Comment

Inspiration

Food waste is a massive problem with over 50 million tonnes of food waste produced every year in Canada alone. 60% of it is avoidable with better planning and awareness, and a majority happens at the household level where food is forgotten, spoiled, or unused.

We were inspired by a simple question: What if your fridge could nudge you to use what you already have? nudge. was born from the idea that a simple reminder can lead to more sustainable everyday habits.

What it does

nudge is an AI-enabled smart fridge system that helps reduce household food waste by turning your fridge into an intelligent assistant.

A camera mounted inside the fridge continuously scans its contents and sends images to a backend server. Using computer vision and AI, nudge:

  • Identifies ingredients inside the fridge
  • Estimates food freshness and ripeness over time
  • Tracks what’s available before it goes bad
  • Generates personalized recipes based on ingredient availability and user preferences

How we built it

Hardware

  • ESP32 + ESP32-CAM mounted inside the fridge
  • Powered by a battery / power bank for flexibility
  • Grove temperature sensor to monitor fridge conditions
  • Arduino setup with Grove connector shield

Backend

  • Python & TypeScript for backend services
  • MongoDB Atlas for storing ingredient data and history
  • Gemini API for: Image recognition of fridge contents, AI-generated, personalized recipe creation, Computer vision pipeline processes images streamed from the fridge camera

Frontend

  • React with Tailwind CSS and JavaScript
  • Web app dashboard showing: Detected ingredients, Freshness estimates, Suggested recipes based on preferences and availability

Challenges we ran into

  • Hardware reliability: Streaming images from inside a cold, enclosed fridge environment while maintaining stable power and connectivity
  • Computer vision accuracy: Handling overlapping items, poor lighting, and partially obscured ingredients
  • System integration: Coordinating hardware, backend processing, AI APIs, and frontend updates in real time
  • Time constraints: Balancing ambitious hardware + AI goals within a hackathon timeframe

Accomplishments that we're proud of

  • Building a fully integrated hardware + AI + web system
  • Successfully streaming camera data from an ESP32-CAM to a backend server
  • Using Gemini API for both image understanding and recipe generation
  • Creating a solution grounded in real, impactful statistics about food waste
  • Designing Nudge to be practical, scalable, and user-friendly—not just a demo

What we learned

  • Hardware-software projects demand strong teamwork and clear division of responsibilities
  • How to build components
  • How to route pages
  • How to use prototyping in Figma

What's next for nudge.

  • Introduce expiry alerts and proactive “use-this-today” nudges
  • Add user profiles with nutrition goals
  • Add a Favourites section to bookmark recipes
  • Improve ingredient recognition accuracy and freshness estimation
  • Explore partnerships with non-profit organizations and restaurants for greater impact
Share this project:

Updates