EcoSort AI : Intelligent Waste Management System

Home screen of EcoSort AI showing user interface where users can type, speak, or upload an image to identify waste and get disposal guidance
Result screen where project identifies a plastic bottle as recyclable and provides disposal guidance, scrap value, and creative reuse ideas.
EcoSort AI identifies a battery as hazardous waste and provides safety guidelines, disposal options, and handling precautions.
Impact dashboard showing user contributions, including recycled, composted, and safely disposed waste along with earned eco badges.
EcoSort AI game lets users quickly sort waste into correct categories, promoting awareness through an interactive experience.
EcoSort AI uses mobile image recognition to detect keyboard as hazardous e-waste and guide safe disposal
Login interface where users can sign in to access personalized features like eco-points, dashboard, and activity tracking.
EcoSort AI home screen featuring text, image, and voice input, allowing users to identify waste and receive AI-based disposal guidance.

🌍 Inspiration

Waste mismanagement is a growing global issue, yet many people still struggle to identify how to dispose of everyday items correctly. This often leads to recyclable or organic waste ending up in landfills, causing pollution and environmental harm.

We were inspired to build a solution that is simple, intelligent, and accessible — one that can guide users instantly using AI, even in low-connectivity environments.

💡 What it does

EcoSort AI is an intelligent waste management system that helps users identify and properly dispose of waste using AI.

Users can:

Type, speak, or upload an image of waste
Instantly classify items as recyclable, organic, hazardous, or general
View decomposition time and disposal tips
Get DIY reuse ideas
See real-world environmental impact warnings
Earn eco-points and track their impact

🛠️ How we built it

We built EcoSort AI using:

HTML, CSS, JavaScript for a responsive and interactive UI
TensorFlow.js + MobileNet for real-time image classification directly in the browser
LocalStorage for storing eco-points, impact data, and user progress
Service Workers to enable offline functionality (PWA support)

The system uses a custom waste database with smart matching logic and aliases to map AI predictions to meaningful waste categories.

⚡ Challenges we ran into

AI Accuracy: MobileNet provides general labels, so mapping predictions to specific waste categories required custom logic
Offline Support: Ensuring smooth performance without internet using caching and service workers
Data Mapping: Building a meaningful waste database with correct categories, tips, and warnings
User Experience: Balancing multiple features (AI, voice, game, dashboard) while keeping the interface simple

🏆 Accomplishments that we're proud of

Successfully implementing on-device AI for waste detection
Building a fully functional PWA that works offline
Creating a multi-feature platform with voice input, image detection, and gamification
Adding real-world environmental awareness through dynamic warnings and impact tracking
Designing an engaging and modern UI with smooth interactions