EcoSort

The days of dumping everything into the garbage can are over. With EcoSort, a single click can tell you all the information you need to dispose of something correctly for the good of the environment.

Comment

Inspiration

Imagine this: you are standing in front of the set of three garbage cans. Because you care about releasing less greenhouse gases to fuel the oven that heats the Earth, you might spent a minute to search up what bin the cheese-powder-covered-aluminum-and-paper-sandwich-wrapping-in-a-plastic-bag should go...and then give up, frustrated at life. But fear not, EcoSort is here to support!

What it does

EcoSort is a web application that works on both mobile and desktop devices with its quick responsiveness that helps you sort your garbage appropriately. If an image is uploaded, EcoSort uses Clarifai's object detection capabilities to identify the item and generate a corresponding text description. It then utilizes OpenAI’s ChatGPT API to analyze the item's description and determine the appropriate bin for disposal. EcoSort then provides clear, easy-to-understand, and visually pleasing information to users, guiding them on whether the item should be recycled, composted, or placed in the landfill bin, as well as giving an environmental safety rating.

How we built it

Our project was developed for the web using React and node.js. We built our front end with ReactJS + TypeScript. This allowed us to have a production ready and type safe front end. We styled our front end with a combination of plain CSS and TailwindCSS. TailwindCSS allows for streamlined and pretty styles. Our back end was built with JavaScript and express.js. We use Clarif.ai and ChatGPT to identify what objects are and know their appropriate disposal methods.

Challenges we ran into

At the start, we struggled to find an accessible API for object detection, first resorting to Teachable Machine. While it was exceedingly easy to train, we had problems with its limited versatility and integration. Therefore, we decided to switch to Clarifai, which can recognize a wider variety of items. Later, we found that ChatGPT API would answer in inconsistent ways, leading to some intensive prompt-engineering. After dozens of iterations, we got the chatbot to give the answer in a specific format. Of course, debugging 404 errors are inevitable along the way.

Accomplishments that we're proud of

We are proud of successfully developing EcoSort within the tight timeframe. Utilizing Clarifai API and the GPT-3 API, two cutting-edge AI technologies and integrating them were two significant achievements for us. That, paired with our user-friendly interface, EcoSort has the potential to steer daily waste management practices in our community towards a greener environment, and we are super excited for that.

What we learned

During the development of EcoSort, we deepened our understanding of object detection and image classification techniques, as well as leveraging powerful language models for intelligent text analysis. We also enhanced our skills in frontend and backend development, API integration, and teamwork. Additionally, working on EcoSort increased our awareness of the environmental challenges caused by improper waste disposal and the importance of promoting sustainable waste management practices.

What's next for EcoSort

In the future, we plan to further refine the object detection algorithm that EcoSort uses in order to make the text label more specific to account for minor differences in the state of the item. In addition, we would like to expand the database of recognized objects to improve waste disposal recommendations. Enhancements such as integrating location-specific recycling guidelines and implementing a feedback system are also part of our future plans. We hope that ultimately, EcoSort can contribute to a more sustainable and eco-conscious society.

P.S. The demo link is front-end only!

Built With

Share this project:

Updates