We are taken aback by the amount of food, whether at an event or even at home, that is wasted everyday. We realized that our habits have made it easier to just throw away food than to provide it to those who cannot provide themselves with meals. Our vision is to fill as many hungry stomachs as possible by making food donation so convenient that food disposal will be out of the question.

What it does

Our app allows users to donate their excess food to food banks through UPS drivers traveling en route. The user takes a picture of the food items and answers a few simple questions to ensure the food's preservability and quality. The app will then let the user know when the UPS driver will arrive to the specified location to pick up the food items.

How we built it

Feed was built on Swift. We utilized Apple's machine learning model Resnet50 to detect the food that the user is trying to donate. We also utilized MapKit to construct an interactive interface so users can view food bank locations and the distances from their location. We also utilized CoreLocation, AVKit, and Vision to get the addresses of the food banks, access the device's video functionalities, and analyze the pictures of food to be donated, respectively. We stored the user and food bank information in a database through MongoDB. Our backend was build on Python, Flask, and Javascript. The optimal food bank was chosen from an all-encompassing algorithm that calculates the shortest distance between the user's location and food banks and takes the food bank's requirements into account.

Challenges we ran into

We had to figure out how to incorporate machine learning to enhance the user's experience. It was our first time working with machine learning models and had to start from scratch.

Accomplishments that we're proud of

We are very proud of our product. We spent a lot of time to make the backend error-proof and the interface visually aesthetic. We are also proud of being able to learn and incorporate new technologies. Most importantly, we are proud to have spend our weekend on such an impactful project.

What we learned

We learned that food banks typically have strict requirements on the food that is donated to them. We had to accommodate these requirements through machine learning to ensure that the food donated will be acceptable.

What's next for Feed

Feed was built on a relatively small scale, taking only the metro Atlanta area into consideration. We hope to make Feed applicable on a national scale. We also hope to make the user's job in requesting a food donation pickup more convenient by incorporation more machine models to determine the food's serving size and preservability (without having the user input it). Most importantly, we hope to fill many more stomachs by expanding our drop off locations.

Built With

Share this project: