FoodSaver

Inspiration

While food wastage might only be brought up during the holidays, the truth is that in America, a large amount of food is thrown away every day. In fact, for every person in the United States, approximately one pound of food is wasted every day. This leads to over 80 billion pounds of food waste being generated in a year. Through FoodSaver, we hope that people can be more conscious about the food they have in their pantry and make an effort to waste less food.

What it does

FoodSaver is a mobile application that has two main services: a camera and logger function.

Camera - FoodSaver prompts the user to take a picture of their produce by tapping a button. After taking the picture and having the user confirm they are ready to move on, the photo is sent to an API.
Logger - FoodSaver retrieves the data that is returned by the API. Based on the data, the logger determines whether to increment the number of fresh or spoiled apples, bananas, and oranges.

How we built it

FoodSaver has three main components: an API, a machine learning model, and a mobile application.

API - FoodSaver uses the Postman development tool to build, test, and modify the API.
Machine Learning Model - FoodSaver uses an image classifier model using TensorFlow, and a pre-trained convolutional neural network (CNN) called Inception. It was trained using a public dataset from Kaggle; however, we modified the dataset to what was best for our application.
Mobile Application - FoodSaver is built on an IDE called Android Studio.

Challenges we ran into

We had difficulty in taking a picture through the mobile application and then sending the photo to the API, integrating all parts of the mobile application. and doing mobile application for the first time, as some members of the team were not familiar with Android Studio.

Accomplishments that we're proud of

We are proud of being able to successfully run the machine learning model on the API, as well as integrating it to the mobile application. We were also satisfied that the API was able to return a response in real time, quickly. More than anything, we were proud of being able to successfully connect machine learning with an android application.

What we learned

We learned how to upload an image from the application to an API and how to use Android Studio to design the application.

What's next for FoodSaver

In the future, we hope to incorporate more functions into the application. For example, in our original plan, we wanted to include aspects such as AR, but because of our limited time, we were unable to get to that. We would also like to improve the model.