Inspiration
Trying to be environmentally conscious can be hard, but it doesn't have to be. We can scan anything you throw away and see if it can be recycled instead!
What it does
It's an app that goes through items within sight of your phone's camera and overlays 3D models representing a particular concept helping to reinforce that item can be recycled. It also roasts the person based on messiness of the area.
How we built it
It was built using Unity, C# and Niantic Lightship. I leveraged AI through Gemini to generate soft roasts based the detected content and GPT-4 to categorize into the type of recycling that could be done based on the classified image type. I briefly tried to add YOLOv4 and a custom camera to texture setup before realizing that Niantic has ported the more powerful YOLOv8 already.
Challenges we ran into
Had to do everything myself. PM work. Slides, Solo UI/UX and extensive testing to ensure it does not fail in the most critical of situations.
Accomplishments that we're proud of
I soloed an entire full stack application that combines Gen-AI, AR, computer vision in the span of 36 hours. It was tough and no-one to lean onto, but provides more fascinating results.
What we learned
How to make the most out of the Calhacks trip while riding solo
What's next for SmartCycle - Your Recycling Companion!
Better classification system and more classes (maybe a custom trained CV model), probably fix up the UI
Log in or sign up for Devpost to join the conversation.