A community-driven litter reporting and cleanup platform built at DeltaHacks 12.

https://streetsweepai.vercel.app/

• Backend: FastAPI (Python), Uvicorn
• Database: MongoDB Atlas
• Storage: Cloudinary (image CDN)
• Auth: JWT (see auth.py), bcrypt for password hashing
• AI: Google Gemini for image classification (see tickets.py classify endpoint)
• Frontend: React/TypeScript CORS enabled on backend
• Deployment: Frontend deployed on Vercel, Backend deployed on Railway

• Toronto traffic camera metadata: https://data.urbandatacentre.ca/catalogue/city-toronto-traffic-cameras
  • Used to locate camera latitude/longitude and intersection name during ingestion (ingest_general.py).
• Test images: local folder testImages/ (ignored in git).

1. Create and activate a venv.
2. pip install -r requirements.txt
3. Set environment variables (see .env.example pattern):
   • MONGO_URI
   • CLOUDINARY_URL
   • GEMINI_API_KEY (for classify endpoint)
4. uvicorn main:app --reload

• Lukhsaan Elankumaran
• Harry Lu
• Varun Gande
• Andrew Law