EcoVenture (FORK OF KS2013 REPO THIS ISNT MINE)

Turn Trash Into Rewards - A cross-platform app that rewards you for cleaning up litter in public places.

Features

• Video Recording - Record yourself picking up and disposing of litter
• Multi-Model AI Detection - Real-time litter detection using:
  • TensorFlow.js COCO-SSD (local, offline)
  • MobileNet for enhanced classification
  • TACO dataset categories for specialized trash detection
• Location Verification - Ensures you're in a public place (park, street, beach, etc.)
• Points System - Earn points for verified litter disposal
• Gift Card Rewards - Redeem points for Amazon, Starbucks, Target, and more
• Organized Cleanups - Join community cleanup events with RSVP and check-in
• Cloud Sync - Supabase backend for user data persistence
• Cross-Platform - Desktop (Windows, Mac, Linux) and Android

Quick Start

Prerequisites

• Node.js 18+
• Android Studio (for Android builds)

Installation

# Install dependencies
npm install

# Run the Electron app
npm start

# Run in dev mode (with DevTools)
npm run dev

Building

# Build for Windows
npm run build:win

# Build for Mac
npm run build:mac

# Build for Linux
npm run build:linux

# Build for all platforms
npm run build

Android

# Sync Capacitor
npm run android:sync

# Open in Android Studio
npm run android:open

# Run on device/emulator
npm run android:run

How It Works

1. Open the app on your device
2. Allow camera and location access
3. Go to a public place (park, beach, street, etc.)
4. Enable live detection to see AI recognizing litter
5. Record yourself picking up litter and disposing of it
6. Submit for verification - AI analyzes your recording
7. Earn points and redeem for gift cards!

Points System

Action	Points
Verified disposal	50 pts
High confidence (>80%)	+20 pts
Multiple items	+10 pts each
Daily streak	+25 pts x days
First submission	+100 pts

Available Rewards

• £/$5 Gift Card - 5000 pts
• £/$10 Gift Card - 9500 pts

Tech Stack

• Desktop: Electron
• Mobile: Capacitor (Android)
• AI/ML: TensorFlow.js COCO-SSD, MobileNet
• Backend: Supabase
• Frontend: Vanilla JS, CSS3, HTML5
• Storage: electron-store (desktop), Supabase (cloud)
• APIs: OpenStreetMap Nominatim for geocoding

Project Structure

ecoventure/
├── main.js              # Electron main process
├── preload.js           # Secure bridge between main and renderer
├── capacitor.config.ts  # Capacitor configuration
├── app/
│   ├── index.html       # Main HTML
│   ├── styles.css       # Responsive styles
│   ├── app.js           # Frontend logic with AI detection
│   ├── supabase.js      # Supabase client
│   └── js/              # Modular JS files
│       ├── auth.js      # Authentication
│       ├── camera.js    # Camera handling
│       ├── cleanups.js  # Organized cleanups
│       ├── detection.js # AI detection
│       ├── friends.js   # Social features
│       ├── leaderboard.js
│       ├── rewards.js   # Rewards system
│       └── ui.js        # UI utilities
├── android/             # Android project (Capacitor)
├── assets/              # App icons
└── dist/                # Built executables

AI Detection

The app uses multiple AI models for robust trash detection:

• COCO-SSD: Detects 80+ object classes including bottles, cups, food items
• MobileNet: Additional classification for confidence boosting
• TACO Categories: 60+ specialized trash categories (plastic bags, cigarettes, cans, etc.)

All detection runs locally on your device - no server required.

License

MIT