RUTrash

Empowering Sustainability: Trash Classification, Carbon Footprint Analysis, and Climate Expert Chat.

Comment 1

Inspiration

At the welcoming ceremony hosted by the Rutgers Office of Climate Action, they introduced their mission and motivated us to contribute positively to environmental conservation. Following the event, we noticed individuals discarding pizza boxes and water bottles into general waste bins, which highlighted a prevalent issue: improper waste disposal. Our research revealed that approximately 75% of waste is recyclable, yet only 30% is correctly recycled, and about 22 million tons of food waste, which could be composted, end up in landfills instead. This waste contributes to greenhouse gas emissions equivalent to those from 2 million cars (stat as of 2015). Recognizing this, we aimed to develop a solution that makes individuals aware of their environmental footprint through their daily actions, thereby encouraging them to reflect on and reduce their impact on the planet.

What it Does

Functionality of the Project

  • Trash Type Detection: Allows the user to upload an image of their waste/trash, determines the type of trash using AI, and provides suggestions on suitable disposal methods.
  • Carbon Footprint Calculation: Enables users to input data about their habits (e.g., electricity usage, mileage) to determine their carbon footprint.
  • Climate Expert Chatbot: Incorporates an AI chatbot that engages users in conversations about climate sustainability and ways to protect the environment.

How We Built It

Technology Used

  • Cloudflare Workers AI API
  • Flask
  • Pytorch
  • Torchvision
  • Python Markdown Library
  • Python
  • HTML
  • CSS

Description

We trained a MobileNetV2 CNN using a dataset from Kaggle on garbage classification. Integrating this model into a Flask app, we developed trash type detection. Additionally, we added carbon footprint calculation using Flask and sourced a function for this purpose. Lastly, we integrated a text generation model from Cloudflare Workers AI API to create the Climate Expert chatbot.

Challenges We Ran Into

  • Model Training: Selecting and training a suitable CNN model within the hackathon's time frame.
  • Cloudflare API Integration: Understanding and implementing API calls for the Cloudflare Workers AI API.

Accomplishments We're Proud Of

Successfully implementing all planned functionalities and adding extra features.

What We Learned

Applied knowledge of CNNs, Flask, HTML, and CSS to develop a functional application.

What's Next for RUTrash

  • Transforming the program into a versatile application for various devices.
  • Improving the UI/UX of the webpage.
  • Enhancing image classification using a more robust CNN model.

Try it out for yourself!

Please follow the directions on the ReadMe to try it out for yourself!

Built With

Share this project:

Updates