EcoSort AI

Inspiration

Incorrect waste segregation leads to contamination and reduced recycling efficiency. We wanted to build a simple AI-powered tool that helps users quickly identify the correct disposal method and encourages better recycling habits.

What it does

EcoSort AI lets users upload or scan a waste item image. Using Google Gemini, it classifies the item, detects contamination, suggests preparation steps, and recommends the correct bin color based on Indian waste rules. It also tracks points and scan history.

How we built it

We built the app using Next.js and Tailwind CSS. Google Gemini 1.5 Flash handles image classification through a serverless API route deployed on Vercel. User stats and history are stored using localStorage.

Challenges we ran into

We faced issues with structured AI responses and dashboard state updates across pages. These were solved through prompt refinement and persistent localStorage-based state management.

Accomplishments that we’re proud of

We successfully integrated multimodal AI into a working web application and built a fully functional, deployable product within limited time.

What we learned

We learned practical AI integration, prompt engineering, serverless deployment, and effective state management in a real-world project setup.

What’s next for EcoSort AI

We plan to add real-time camera scanning, barcode support, and region-specific waste rules. Future improvements include cloud-based user accounts, leaderboards, and improved AI accuracy through feedback and data refinement.

Built With

• next.js
• react
• vercel