https://streamable.com/9fr2e7 (Website Demo)

Smarbin

Inspiration

Recycling contamination is a $200 billion global problem - with millions of tons of recyclables ending up in landfills due to improper sorting. We created Smarbin to bridge the gap between good intentions and proper recycling action, using technology to make sustainability effortless.

What it does

Smarbin is an intelligent waste bin that:

Automatically sorts recyclables (plastic, paper, metal, etc.) from trash using real-time AI classification
Tracks recycling habits through a companion mobile app
Reduces contamination in recycling
Provides rewards for users who recycle

How we built it

Hardware

Arduino with camera and motors
Servo-controlled sorting mechanism

Software

Sklearn model trained on 4,000+ waste images
Custom computer vision pipeline for real-time classification
React + vite app
Django local backend database for RESTful API
Supabase web server backend for user auth + db

Challenges we ran into

Classification accuracy - Early versions struggled with:
- Similar-looking materials (plastic vs. waxed paper)
- Crushed/obscured items Solution: Augmented training data with synthetic images
Hardware constraints:
- Servo jamming with heavy items
- Camera focus in low light Solution: Added IR illumination and torque-resistant actuators
User experience:
- Balancing speed vs accuracy
- Making the system intuitive Solution: Added audio cues and progressive feedback

Accomplishments we're proud of

✅ 92% classification accuracy on common waste items
✅ <1 second processing time per item
✅ Designed for <$100 manufacturing cost

What we learned

The importance of edge computing for real-time applications
How small design choices (like bin opening size) dramatically affect user behavior
That sustainability tech must be both accurate AND convenient to drive adoption

What's next for Smarbin

🔄 Phase 1 (6 months): Campus pilot program with 50 units
📊 Phase 2: Enterprise analytics dashboard for waste management
🌱 Phase 3: Add more bins + make new dropper prototype for more recycling types to further improve sorting

🏭 Long-term: Partner with municipalities for smart city integration