Inspiration
We’ve all stood in front of a recycling bin with a coffee cup or a greasy pizza box, totally unsure of where it goes. This confusion leads to "wishcycling", tossing non-recyclables into the blue bin, which actually contaminates the whole batch and makes the process fail. We wanted to build a tool that replaces that guess.
What it does
EcoSort is a mobile-first web app that uses your camera to identify waste in real-time. By scanning an object, the app tells you exactly how to dispose of it—whether it belongs in the Recycle, Landfill, or Compost bin.
How We built it
We used TensorFlow and Teachable Machine to create a custom image classification model. We used HTML, CSS, and JavaScript for the frontend.
Challenges We ran into
Building a functional AI was difficult because "waste" has no standard shape. A plastic bottle could be clear or crushed, and the AI needs to be able to distinguish it from the user's hand.
Accomplishments that I'm proud of
We are proud of successfully implementing a real-time computer vision pipeline that runs entirely in the browser. Getting a custom-trained model to accurately identify crumpled, real-world objects was a major milestone.
What We learned
We learned that data quality beats data quantity. Five photos of a bottle in different lighting are worth more than fifty identical shots.
Built With
- css
- firebase
- githubpages
- googleteachablemachine
- html
- javascript
- tensorflow
Log in or sign up for Devpost to join the conversation.