Fridge Friend

Fridge Friend makes cooking your next meal convenient, sustainable, and delicious. Upload an image of your fridge and get recommendations of real recipes that match your fridge best!

Comment

Inspiration

  • We were discussing our lives recently and realized we were often too busy to cook meals for ourselves
  • Realized the importance of being able to cook meals out of ingredients we already have

What it does

  • Users can upload an image of their fridge and the detected items will be generated
  • The user can then edit the ingredients if any mistakes were made, or add other ingredients they did not take pictures of
  • A list of the most closely related recipes will be recommended to the user, including images, ingredients, and instructions

How we built it

  • Python
  • Streamlit with CSS/HTML for the front-end and for image recognition
  • Recipe dataset from Kaggle, vectorizer trained using Scikit-learn
  • PropelAuth for user authentication system
  • MongoDB Atlas for storing user data

Challenges we ran into

  • Found it difficult to use version control with our larger files, but found ways to make it work
  • Learning how to use Streamlit under tight time constraints

Accomplishments that we're proud of

  • Creating a functional application!
  • Login system
  • Using Streamlit to create an organized front-end
  • Recommendations algorithm

What we learned

  • How to use Streamlit
  • A lot of things about Git

What's next for Fridge Friend

  • Improvements on image recognition model - it currently is not very precise at detecting images with many food items, which is important to the functionality of being able to take a picture of your fridge
  • We'd like to be able to make our recommendation system faster
  • Allow accounts to share pictures of their meals with each other
  • Gamify the process to provide incentives for users to make meals

Built With

Share this project:

Updates