GreenSweep

Inspiration

San Francisco faces an 18,000+ ton street trash problem every year, much of which ends up in the Bay and ocean. We wanted to empower citizens to help — using AI + gamification to turn cleanup into a fun and impactful community activity.

What it does

GreenSweep lets users snap a photo, detects trash using AI, provides smart cleanup tips, and tracks verified cleanups on a leaderboard. It transforms civic cleanup into an engaging, measurable game.

How we built it

We combined a custom-trained trash detection model, AI-generated tips, and a responsive web app with camera support. The focus was on creating an experience that is simple, fast, and fun for users.

Challenges we ran into

We tackled camera access & permissions across devices, tuned AI detection accuracy, and optimized the flow to be seamless and rewarding. Balancing real-world variability in street trash photos was also key.

Accomplishments that we're proud of

We built a fully functional, live AI-powered app that works today — with real trash detection, GPT-powered tips, and gamified progress tracking. It’s already ready to help clean real city streets.

What we learned

We learned how to make AI useful and accessible for everyday civic action, and how gamification can motivate people to contribute regularly. Also, how small, local actions can scale up to big impact.

What's next for GreenSweep

We plan to add more functionality (like progress tracking & badges), integrate crowdsourced cleanup data, polish the website UX, and improve the AI model accuracy. Our next goal is to make the app more engaging and more effective for real-world civic use.