The Sustain app

Have you ever wanted to contribute to the Earth's health? The sustain app is for you! With a interactive and competitive set of weekly green goals, you can challenge friends and track your progress!

Comment

Inspiration

The point of the sustain app is to bring a competitive spirit and a rewarding feeling for doing good actions that help the environment. The app is basically a social media app where you can add friends, and see your leaderboards and community progress toward a green goal you set for yourself. The intended way to use the app is that every time you find a can or a piece of garbage on the ground, you can scan the item using the machine learning algorithm that over time will be able to detect more and more garbage of all different types and then you throw it away to get points(based on garbage type) that stack up over the weeks. The app also keeps track of the barcode so that it isn't used over and over again to hack points. We also planned in the future to add a variety of other methods to gain points such as ride sharing or using reusable containers or bottles.

How we built it

We built the front end using HTML and CSS along with frameworks like Tailwind or Bootstrap. For the back end, we created it using Django. Finally, the machine learning part was implemented using PyTorch and the YOLOv5 Algorithm.

Challenges we ran into

We encountered several challenges in deploying an object detection algorithm that was both accurate and lightweight, ensuring it did not significantly impact inference time which would hurt user experience. This required us to curate a dataset that's diverse while also having a small model. Additionally, we faced significant difficulties in implementing a conversational virtual assistant using AWS for our system. Despite investing more than three hours in setup, it ultimately crashed, leading us to unfortunately drop the idea.

What we learned

For the 2 front-end developers, we learned a lot about how HTML works and the different available frameworks. Our back-end and machine-learning developers learned how complicated it could be to implement chatbots and deploy them onto web apps. They have also been able to learn new technologies such as YOLOv5 and frameworks like PyTorch and Ultralytics.

What's next for the Sustain app

Built With

Share this project:

Updates