-
Home Page
-
Recycle Result
-
Compostable Result
Inspiration
With a high variety of different materials, it is often confusing for us students, and many other people, to determine which bin to throw their trash in.
What it does
Our project is a web application that uses image recognition to instantly classify waste as recyclable, compostable, or trash. Users can upload a photo of their waste item, and our platform provides real-time sorting guidance to ensure the item ends up in the right bin.
How we built it
We started off experimenting with Tensorflow, Sklearn, and several similar libraries when making the artificial intelligence brain of the app. However, after a lot of experimentation and solid AI models, we decided to use Google CLIP with Torch for the final product. We did our front end in Flask, HTML, and CSS after iterating through other frameworks like Flutter and Firebase. Lastly, we integrated our backend and frontend through Flask.
Challenges we ran into
We ran into several challenges in making this app. Firstly, making the artificial intelligence itself was really difficult. We were initially collecting tons of training data, using libraries that we didn't have much experience with like TensorFlow and sklearn, and we eventually got to make an AI model that was 95% accurate in sorting images across six categories that were then divided into compost, landfill, and recycling. However, this wasn’t good enough for us so we ended up integrating Google CLIP along with the other to get the final creation. The second main challenge we had was building the front end and integrating it with Flask. Through a lot of hard work and perseverance, we ended up with a final product that we are proud of.
Accomplishments that we're proud of
We are proud of what we accomplished in the short amount of time that was allocated for this project. We are very proud of our ability to train our AI to detect trash and recognize various materials. We are also very proud of the full-stack platform that we were able to create. This platform has an interactive page that allows for media upload which gets scanned and displays exactly what type of trash it is.
What we learned
What's next for Trash Buddy
Our vision for this project is much more than what was created during this hackathon. We have plans to increase the accuracy of our trash detection AI and create an app that allows for trash recognition on the go. We also will focus on creating a kids-friendly platform that is interactive and allows kids to learn about trash and the importance of recycling.
Log in or sign up for Devpost to join the conversation.