Re:Cycle

Inspiration

The world is experiencing one of the largest climate shifts in its history. We noticed that many people wanted to be more environmentally friendly, but did not know where to start. We also noticed that many people did not know where to dispose of their items. Most items end up in the wrong bin, which has huge consequences on waste stream treatment. To solve this problem, we wanted to create an app that would tell users where their trash should go. Overall, we hope that this app will encourage more environmentally sustainable practices that will benefit municipal waste management systems, saving billions.

What it does

Re:Cycle is an app that utilizes AI to categorize waste products. Users can choose between taking a photo with the in-app camera, choosing an image from their gallery, or searching on the app database for a specific item or product. The app will then return with a suggestion for where to dispose of the product.

How we built it

Our team was divided into two groups: front-end and back-end developers. Front-end developers started with wire-framing and prototyping using Adobe XD, then implemented using Flutter. Back-end developers utilized Google Cloud’s Vision API for image identification/classification and NodeJS for image server.

Challenges we ran into

Initially, our biggest problem was getting started. We had a great idea and many proposed solutions, but we were limited by the capabilities we had as a group of making any feasible app. A large chunk of time was dedicated to training Google Cloud’s vision API to return accurate identifiers for given images. Another issue we struggled with was the fact that no one was familiar with either Flutter as a framework or Dart as a language.

Accomplishments that we're proud of

We are proud of how far we got with Flutter. It was something that none of us were even remotely familiar with, but we eventually got something more than the yellow and black error screen. We are also happy that we were able to implement Google’s Vision API. The model was trained with over 2,500 images to classify the different waste items.

What we learned

We learned that having a good team and a great idea makes it easy to have a structured plan. For some of us, this was our first hackathon. For others, this was one of the first times we ever tried creating an app. The process has opened our eyes to the complexities that arise in app development.

What's next for Re:Cycle

We need to continue building out our user interface—that includes having a built-in database so that users can search for items in case of AI identification failure. We also need to train the AI to identify more than just a single object in an image. Finally, having thorough user recommendations for item disposal is necessary, particularly for common, difficult to recycle items like makeup or electronics. We are open to feedback as well! Please comment below any suggestions for improvement.