Description as a Tweet:

94% of Americans support recycling, and 74% say it should be a top priority, but only 35% of people recycle. The reasons behind it are sorting & convenience. With just a click, RecycleIT will help you identify your recyclable products & give you a convenient location to recycle.

Inspiration:

Our generation should do more and limit the negative impact of not recycling. The benefits of recycling are not just limited to the environment. There are economic and social advantages of recycling as well. Recycling conserves energy, reduces air and water pollution, reduces greenhouse gasses, conserves natural resources, and creates jobs and tax revenue.

What it does:

Our application predicts whether a product is recyclable or not. When you take a picture on your phone camera, it tells you what that object is in real-time and if you can recycle it. Our application can now detect five classes/Objects - boxes, plastic bottles, glass bottles, cans, crushed soda cans, and soda cans. We plan to add more classes and give the location of the nearest recycling center.

How we built it:

We build the android application using React Native. We used open-source data and analyzed the data, and pre-processed it. We trained our own model using Convolution Neural Network in python for image classification. Finally, we integrated the CNN model with the android application using TensorFlowJS.

Technologies we used:

  • HTML/CSS
  • Javascript
  • Node.js
  • React
  • Python
  • Misc

Challenges we ran into:

Due to Pandemic since the HackHer413 was virtual, the biggest challenge we faced was to have all the software and tools installed correctly in all the team member’s systems. It was essential to have a common environment so that we have an accurate and thorough unit and integration testing of our app. Also, another challenge we faced was to train and deploy the classifier to accurately categorize the wastes.

Accomplishments we're proud of:

We are proud that we could showcase our skills through the project. Our trained model has an accuracy of 89.53%, and we successfully deployed the trained model in our application.

What we've learned:

We learned new skills from our team members like React Native, TensorFlow, and Expo. We also learned how to collaborate under pressure.

What's next:

Our application will provide the nearest location to the recycling center from the user. We want to add more categories and more types of images to our trained model. We want our application to be able to distinguish the waste among organic waste, recyclable waste, non-recyclable waste, E-waste, etc. We also want to add instructions on how to recycle an item properly.

Built with:

Design Tools: Figma, Canva
Dataset: Open Souce Data - http://web.cecs.pdx.edu/~singh/rcyc-web/dataset.html
Software Tools: VSCode, Google Colab
Technologies: React Native, Javascript, ReactJS, Python, TensorFlow, Expo

Prizes we're going for:

  • Fujifilm Instax Mini 11 Instant Film Camera, Sky Blue
  • Cash prizes: $1,000 total to first place team; $500 total to second place team; $200 to third place team
  • Hydroflasks
  • Apple AirPods 3rd Gen
  • JBL Clip Bluetooth Speakers
  • $500 total to the winning team
  • Meta Portals

Team Members

Vaishali Mahipal
Riddhi Damani
Vadim Egorov

Table Number

Table TBD