MLPrep

Get recipes with just a picture of ingredients! 🍝🥗

Comment 1

🥗 What is MLPrep?

MLPrep (pronounced MealPrep) is a mobile app that offers recipe recommendations based on a picture of the ingredients, aimed to help reduce food waste by helping users figure out a meal that maximize the usage of their available ingredients 🥗.

Using machine learning, MLPrep eliminates the manual work and allows users to simply take a picture of the ingredients they want to cook with 📷, and our app will automatically recognize the ingredients in the picture and generate recipes accordingly.

Clicking on each recipe brings the user to its site with detailed preparation and cooking instructions. Users can also save recipes into their favorites list for easy and convenient access in the future.

💡 Inspiration

During the pandemic, a lot more people are buying groceries and cooking at home 👩‍🍳👨‍🍳 (us included!). Homemade food is awesome, but the food being wasted as a result isn't. Individual households are responsible for the largest portion of all food waste, with fresh fruits and vegetables account for the largest losses at the consumer level (Source: https://foodprint.org/). 2/3 of food waste can be attributed to food spoilage, which often is caused by improper storage or partially used ingredients.

We want to help reduce food waste by providing an fast and easy way for new "chefs" to discover recipes and utilize the most of their available ingredients before they spoil. Let's all cook sustainably 💚!

🔨 How we built it

MLPrep is a Flutter mobile app, integrated with a Tensorflow-lite model and Flask backend. We decided to develop on the mobile platform in order to make it convenient for users to utilize their phone's camera functionality to take pictures of ingredients. We chose Flutter as our development platform due to its cross-platform capability, which allows our app to be more accessible to more users.

🤔 Challenges we ran into

  • Limited capabilities of virtual emulators (the camera functionality can't be tested through an emulator and some of our members couldn't run the app on an ios phone, so testing was difficult)
  • Integrating our custom Tensorflow model with the Flutter frontend (our model was able to classify individual images of ingredients, but not multiple ingredients in one photo, so we ended up going with the SSD model)
  • Flutter learning curve

💪 Accomplishments that we're proud of

  • Our first full-fletched Flutter mobile app!
  • This is our first time integrating machine learning into a project, and we're proud that we have a working MVP

🎯 What's next for MLPrep

  • Allow users to sort/filter through recipes based on certain criteria or preferences (allgeries, diets, preferred cuisines, etc...)
  • More robust model that allow for greater variety of ingredients classification

Built by HackSparrows 🐤 (Pod 1.1.2)

  • Chau Vu
  • Emily Amspoker
  • Mondale Felix

Built With

Share this project:

Updates