EcoSan

🌱 Inspiration

Improper waste disposal is one of the most common yet overlooked contributors to pollution and climate change. Many people want to recycle or dispose of waste responsibly but often don’t know which bin an item belongs to, how harmful it is, or what better alternatives exist.

We were inspired by this everyday confusion and asked a simple question: What if people could get instant, intelligent guidance at the exact moment they throw something away? EcoScan was created to turn that moment into an opportunity for learning and positive climate action.

♻️ What it does

EcoScan is an AI-powered waste sorting and sustainability tracking app that helps users:

Scan waste items using their phone camera

Identify the waste type using image recognition

Get clear instructions on how to dispose of it correctly

Learn the environmental impact of their choices

Discover eco-friendly alternatives

Track how much carbon, water, and waste they have saved

Compete on a global leaderboard with other users based on environmental impact

By combining education, AI, and gamification, EcoScan makes sustainable behavior simple, engaging, and measurable.

🛠️ How we built it

Frontend: Built as a Flutter mobile app with smooth animations, light/dark mode, and a clean, student-friendly UI.

AI Vision: Used Teachable Machine to train an image classification model that identifies common waste categories such as plastic, paper, glass, metal, organic waste, and e-waste.

AI Decision Engine: Integrated Gemini 2.5 Flash to generate intelligent disposal methods, environmental impact explanations, and sustainable alternatives.

Backend: Used Supabase for authentication, database storage, and real-time leaderboard data.

Authentication: Implemented secure email-and-password login using Supabase Auth.

Impact Tracking: Each scan estimates carbon saved, water saved, and waste diverted from landfills, which contributes to user eco-points and leaderboard rankings.

🚧 Challenges we ran into

Training an image classification model that works reliably in real-world conditions with different lighting and backgrounds

Balancing model accuracy while keeping the number of waste categories manageable

Designing a leaderboard system that encourages competition without discouraging new users

Translating environmental data into values that are simple, understandable, and meaningful

Ensuring smooth performance while integrating AI responses in real time

🏆 Accomplishments that we're proud of

Building a complete, end-to-end AI-powered sustainability app within a short hackathon timeline

Successfully combining computer vision and generative AI in a practical, real-world use case

Creating a system that not only educates users but also quantifies environmental impact

Designing an engaging dashboard and leaderboard that motivates long-term sustainable habits

Making the app beginner-friendly while still technically impressive

📚 What we learned

AI can be most impactful when it solves simple, everyday problems

Limiting model complexity can significantly improve real-world accuracy

Good UX is just as important as good technology for behavior change

Sustainability education becomes far more effective when paired with feedback and progress tracking

Collaboration between frontend, backend, and AI systems is key to building meaningful applications

🚀 What's next for EcoScan

Add region-specific disposal rules based on location

Expand waste categories and improve model accuracy

Introduce barcode scanning for packaged products

Partner with schools and eco-clubs for community challenges

Provide detailed monthly environmental impact reports

Enable offline scanning for low-connectivity areas

EcoScan has the potential to grow beyond a hackathon project into a tool that helps people around the world make smarter, more sustainable choices every day. 🌍♻️

Built With

• android-studio
• api
• dart
• flutter
• gemni
• google-source
• machine-learning
• supabse