🌍 CleanCity Agent

📸 Step 1: Location & Image Upload

Start by specifying where you found the trash, then upload a photo for AI analysis.

📊 Step 2: Analysis Results

🤖 Step 3: Automated Reporting & Integration

Real-time automation status will appear here after analysis...

CleanCity automatically integrates with your systems:

• 🔌 Database logging
• 📊 Analytics updates
• 🔔 Crew notifications
• 💼 Cost calculations

🌍 Share Your Impact

Help spread awareness and inspire others to take action! Share your cleanup efforts on social media.

📖 How to Use CleanCity Agent

Quick Start Guide

Step 1 – Specify Location 📍
Enter where you found the trash (street name, park, beach, etc.) or click "Get GPS" to automatically detect your current location using your device's GPS.

Why location matters: Tracking locations helps identify "hotspots" - areas with recurring trash problems that need special attention.

Step 2 – Upload a Photo 📸
Take a clear photo of the littered area and upload it. You can:

• Upload an existing image from your device
• Use your camera to take a photo right now

Photo Tips:

• ✅ Good lighting helps detection accuracy
• ✅ Get close enough to see individual items
• ✅ Include surrounding context (park bench, street sign, etc.)
• ❌ Avoid blurry or dark images

Step 3 – Add Notes (Optional) 📝
Provide additional context like:

• "Near the playground entrance"
• "Behind the main parking lot"
• "Recurring problem - third time this month"

Step 4 – Start AI Analysis 🚀
Click "Start AI Analysis" and watch the magic happen! Our computer vision AI will:

• 🔍 Detect and identify trash items (bottles, bags, wrappers, etc.)
• 📦 Draw bounding boxes around each item
• 📊 Count items and categorize them
• ⚡ Calculate confidence scores

Step 5 – Review Results 📊
You'll see three key outputs:

1. Items Detected 🔍

   • List of all trash items found
   • Confidence scores for each detection
   • Total count across categories

2. Cleanup Action Plan 📋

   • Severity level (Low/Medium/High)
   • Number of volunteers needed
   • Estimated cleanup time
   • Required equipment list
   • Environmental impact assessment

3. Email Report 📧

   • Professional report ready to send
   • Copy and paste to email
   • Share with city officials or cleanup groups

Step 6 – Take Action 🎯
Use your results to:

• ✉️ Report to local authorities
• 👥 Organize a community cleanup
• 📈 Track progress over time
• 🔥 Identify hotspots that need attention

💡 Pro Tips

For Better Detection:

• Take photos in daylight when possible
• Capture multiple angles if trash is spread out
• Focus on one area at a time for accurate counts

For Better Tracking:

• Always add location information
• Be consistent with location names
• Check the History tab to see patterns
• Use the Hotspot Analysis to prioritize efforts

For Community Impact:

• Save all events to build a data history
• Share reports with local government
• Use data to request more trash bins or cleanups
• Document improvements to show success

⚠️ Important Notes

Accuracy: This AI is trained on common litter types but may miss items in poor lighting or unusual positions. Always verify results visually.

Privacy: Images and data are processed locally. No personal information is uploaded except for optional LLM enhancement (if configured).

Purpose: This tool is designed to empower community action, not replace professional waste management systems.

❓ Frequently Asked Questions

🤖 About the AI Detection

Q: How accurate is the trash detection?
A: The AI uses a YOLOv8 computer vision model trained on real trash images. Accuracy depends on:

• Image quality and lighting (daylight is best)
• Camera angle and distance
• Trash visibility (not hidden or buried)

Typical accuracy: 75-90% for common items like bottles, bags, and wrappers.

Q: What types of trash can it detect?
A: Common litter categories including:

• Plastic bottles and containers
• Food wrappers and packaging
• Plastic bags
• Cigarette butts
• Cans and metal items
• Paper and cardboard
• General debris

Q: Why are some items missed?
A: The AI may miss items that are:

• Partially hidden or buried
• In shadows or poor lighting
• Very small (like tiny pieces)
• Unusual or rare trash types not in training data

🔒 Privacy & Data

Q: Is my data stored or shared?
A:

• Images are processed locally on the server
• Event data is stored in a local SQLite database (if you choose "Save to history")
• No images or personal info are sent to third parties
• Optional: LLM API calls (if configured) send text summaries only, not images

Q: Do I need an account?
A: No! This is a free, open tool. No login required.

📊 Using the Results

Q: How should I use the cleanup plan?
A: The plan provides realistic estimates based on:

• Number of items detected
• Types of trash (some require special handling)
• Typical volunteer efficiency

Use it to:

• Request appropriate resources from authorities
• Plan community cleanup events
• Estimate budget for equipment/disposal

Q: Can I edit the report before sending?
A: Yes! Copy the text from the report box and edit it in your email client or word processor before sending.

Q: Who should I send the report to?
A: Consider sending to:

• City/town environmental department
• Parks and recreation department
• Local waste management authority
• Community cleanup organizations
• Neighborhood associations

📍 Location & Tracking

Q: Why does location matter?
A: Location tracking helps:

• Identify "hotspots" with recurring problems
• Show patterns to authorities
• Prioritize areas needing attention
• Demonstrate impact over time

Q: Is GPS required?
A: No, location is optional but recommended. You can:

• Use GPS auto-detect
• Type location manually
• Leave it blank (less useful for tracking)

🚀 Advanced Usage

Q: Can I use this for large-scale city monitoring?
A: This is a prototype designed for:

• Community activists and groups
• Individual concerned citizens
• Small-to-medium cleanup organizations

For large-scale deployment, consider:

• Integrating with city GIS systems
• Adding user authentication
• Cloud hosting for multi-user access
• Professional model fine-tuning for your area

Q: Can I improve the AI detection?
A: Yes! The model file is at Weights/best.pt. You can:

• Train on your own trash images
• Fine-tune for specific trash types in your area
• Replace with a different YOLO model

Q: Can I run this offline?
A: Partially:

• ✅ Trash detection works offline (local AI model)
• ✅ Cleanup planning works offline
• ❌ LLM-enhanced reports require API access
• ❌ GPS reverse geocoding requires internet

Set LLM_PROVIDER=offline in your .env file for full offline mode.

🛠️ Troubleshooting

Q: The detection is very slow. Why?
A: Computer vision is computationally intensive. Speed depends on:

• Your hardware (GPU is faster than CPU)
• Image size (larger images take longer)
• Server load

Typical processing: 2-10 seconds per image.

Q: "No trash detected" but I can see trash in the image?
A: Try:

• Taking a clearer, better-lit photo
• Getting closer to the trash
• Ensuring items are visible (not hidden)
• Adjusting camera angle

If issues persist, the model may need retraining on similar images.

Q: How do I report a bug or suggest a feature?
A: Check the project repository for:

• Issue tracker
• Contribution guidelines
• Contact information

🌍 Making an Impact

Q: Does this really help clean up trash?
A: This tool provides:

• Documentation for authorities
• Evidence of recurring problems
• Data to support cleanup requests
• Organization for community action

Real cleanup requires human action, but this tool makes that action more effective and data-driven!

Q: Can I contribute to this project?
A: Yes! This is an open-source hackathon project. Ways to contribute:

• Test and report bugs
• Suggest improvements
• Share success stories
• Help improve the AI model
• Translate to other languages

Q: How can I share my success stories?
A: We'd love to hear how you're using CleanCity Agent! Share:

• Before/after photos of cleaned areas
• Data insights from your tracking
• Community impact stories
• Tips for other users

📜 View Past Trash Detection Events

Track all your saved trash detection events to identify patterns and monitor progress over time. Use filters to narrow down specific locations, timeframes, or severity levels.

📸 Example Use Cases

See how CleanCity Agent can make a difference in your community!

🎓 Training Resources

Want to learn more about community environmental action?

• 🌍 EPA Community Cleanup Guide
• ♻️ Ocean Conservancy Cleanup Resources
• 🏙️ Keep America Beautiful
• 👥 Community Organizing Best Practices

Start small, document everything, and watch your impact grow!

🤖 Ask Questions or Get Help

Chat with the CleanCity AI Assistant to get personalized advice and answers about:

• 🧹 Cleanup strategies and best practices
• 📊 Interpreting your detection results
• 🌍 Environmental impact and regulations
• 👥 Organizing community cleanup events
• 📧 How to communicate with authorities

Example questions:

• "How many volunteers do I need for 50 plastic bottles?"
• "What's the best time of day to organize a beach cleanup?"
• "How do I convince my city council to add more trash bins?"
• "What equipment is essential for a park cleanup?"

💡 Tip: The more specific your question, the better the advice!