cleanwatch-ai

AI-powered garbage detection for cleaner and smarter cities.

Comment 1

🌍 CleanWatch AI

An AI-powered smart city platform that helps citizens report garbage and enables municipalities to identify sanitation problems using multimodal AI and cloud infrastructure.

🌍 Inspiration

Urban waste management remains a persistent challenge in many cities. Garbage often accumulates in public areas because citizens do not have an easy way to report issues, and municipal authorities lack real-time visibility of sanitation problems.

The inspiration behind CleanWatch AI was to build a simple yet powerful platform where citizens can report garbage by taking a photo, and AI automatically analyzes the image to determine the severity and type of waste.

By combining multimodal AI with cloud infrastructure, the goal is to transform raw images into actionable insights that help municipalities maintain cleaner cities.

⚙️ What it does

CleanWatch AI enables citizens to:

🔹 Capture an image of garbage using a mobile or web browser 🔹 Submit a complaint with location data 🔹 Automatically analyze the image using AI

The system then generates:

🧠 Waste type classification 📊 Severity level 📝 Description of the issue 🧹 Recommended cleanup action

All complaints are visualized on an interactive map showing:

📍 Garbage locations 🔥 Heatmaps of waste concentration 📊 Top garbage hotspots

This allows municipal authorities to prioritize cleanup operations efficiently.

🛠 How we built it

The application is built using a serverless architecture on AWS, enabling scalability, reliability, and cost efficiency.

💻 Frontend

The web interface is built using React and hosted on:

☁️ Amazon S3 🌐 Amazon CloudFront (for global content delivery)

Users can capture images directly from their device camera and submit complaints.

🔗 Backend API

The application communicates with backend services using:

🚪 Amazon API Gateway

This acts as the entry point for processing complaint submissions.

🤖 AI Analysis

Images are analyzed using:

🧠 Amazon Bedrock ⚡ Powered by Amazon Nova 2 Lite

The AI model performs multimodal understanding, extracting structured insights from waste images.

🗄 Data Storage

Complaint data and results are stored in:

📊 Amazon DynamoDB

Images and related assets are stored in:

🗂 Amazon S3

📊 Visualization

The platform displays data through:

🗺 Interactive complaint map 🔥 Heatmap visualization 📍 Garbage hotspot rankings

These insights help city administrators quickly identify high-priority waste areas.

📐 Mathematical Insight

Garbage hotspots can be identified using spatial clustering principles.

The density of complaints in a region can be estimated using:

[ Density = {NumberofComplaints}/{Area} ]

Where:

📈 Higher density indicates a garbage hotspot 🚨 Areas with high density require urgent cleanup actions

🚧 Challenges we ran into

During development, several technical challenges were encountered:

📷 1. Camera Access in Web Browsers

Accessing the device camera required handling permissions and ensuring compatibility across mobile and desktop browsers.

🗺 2. Map Marker Rendering

After deployment, map markers initially failed to appear due to missing Leaflet assets in the production build.

☁️ 3. CloudFront Cache Issues

Updates deployed to Amazon S3 were not immediately visible due to CloudFront caching, requiring manual cache invalidation.

🤖 4. AI Prompt Engineering

Generating structured results from the AI model required carefully designed prompts to ensure consistent outputs such as:

  • Severity levels
  • Waste type
  • Recommended cleanup actions

🏆 Accomplishments that we're proud of

One of the biggest accomplishments of this project was successfully building a complete AI-powered smart city application using a simple and scalable serverless architecture.

CleanWatch AI demonstrates how modern cloud technologies and foundation models can transform a simple citizen report into actionable environmental insights.

🤖 Multimodal AI Integration

By using Amazon Nova 2 Lite through Amazon Bedrock, the system can analyze waste images and automatically generate structured outputs such as:

📊 Waste type ⚠️ Severity level 📝 Issue description 🧹 Recommended cleanup actions

🗺 Geospatial Intelligence

Another major accomplishment was implementing geospatial visualization.

Complaints submitted by users are displayed on an interactive map featuring:

🔥 Heatmaps 📍 Garbage hotspots 📊 Waste concentration analysis

This helps municipal authorities make data-driven decisions.

☁️ Serverless Cloud Architecture

The system is built using AWS services including:

  • Amazon S3
  • Amazon CloudFront
  • Amazon API Gateway
  • Amazon DynamoDB

This architecture ensures:

✔ Scalability ✔ Cost efficiency ✔ Easy deployment

Finally, we are proud that a relatively small amount of code can create a meaningful AI-driven smart city solution.

📚 What we learned

This project provided valuable insights into:

🧠 Building serverless AI applications 🤖 Integrating multimodal foundation models into real-world workflows 🏙 Designing smart city solutions using Amazon Nova 2 Lite 📊 Visualizing geospatial data with heatmaps and hotspot analysis

It also demonstrated how modern cloud services enable powerful applications to be built with relatively small amounts of code.

🚀 What's next for CleanWatch AI

Future enhancements could include:

📱 Mobile application for faster citizen reporting 🔮 AI-based prediction of garbage hotspots 🏛 Integration with municipal task management systems 📢 Automated alerts for sanitation teams

These improvements could make CleanWatch AI an even more impactful smart city solution.

🔗 Project Details

💻 GitHub Repository

https://github.com/jayesh-sonawane/cleanwatch-ai

🌐 Live Application

https://d3o7zmc325a69i.cloudfront.net/

Login credentials:

👤 Username: test 🔑 Password: test

🎥 YouTube Demo

https://www.youtube.com/watch?v=0wF8XO7kXVk

💡 Final Thoughts

CleanWatch AI demonstrates how multimodal AI models combined with cloud infrastructure can empower citizens and municipalities to build cleaner, smarter cities.

With technologies like Amazon Nova and Amazon Bedrock, it is now possible to transform simple images into powerful insights that improve urban environments.

⚡ #Amazon Nova Powered Smart City Solution

Built With

Share this project:

Updates