Project Inspiration Rapid urbanization leads to inefficient waste management: garbage trucks often visit empty dumpsters while overflowing ones pollute urban environments. Aligned with UN Sustainable Development Goal 11 (Sustainable Cities and Communities), SmartWaste AI was created to transform municipal waste collection into a data-driven, automated workflow using Google’s advanced multimodal AI capabilities.
What it does SmartWaste AI bridges IoT logistics and multimodal generative AI to modernize urban waste handling:
Gemini Vision Analysis: Municipal workers or citizens upload image snapshots of waste containers. Gemini 2.5 Flash analyzes fill-level percentages, classifies waste types, and generates immediate status recommendations.
Dynamic Route Optimization: Algorithms automatically filter urgent containers (>70% capacity) and compute the shortest sequential route for collection trucks, significantly reducing urban traffic congestion and fuel usage.
Smart Dashboard: Provides city managers with live telemetry, operational priorities, and visual status indicators.
How we built it
AI Core: Integrated Google Gemini API (google-genai SDK) using gemini-2.5-flash for multimodal computer vision and structured recommendations.
Backend: Built with Python and Flask to handle endpoint API calls, asynchronous routing logic, and dynamic data delivery.
Frontend: Responsive UI crafted with HTML5, CSS3, JavaScript (Fetch API), and Bootstrap 5 for fast visual feedback.
Deployment: Hosted live using Python server tunnels for quick global testing during the hackathon.
Challenges we ran into
Fine-tuning real-time prompt engineering for Gemini Vision to ensure accurate fill-level estimation under varying image quality and angles.
Handling cross-origin requests (CORS) and seamless frontend integration with Flask API endpoints under strict submission time constraints.
Accomplishments that we're proud of
Built a fully functional end-to-end MVP within the hackathon timeframe.
Successfully connected Gemini API for automated visual inspection with real-time dynamic route recalculation.
What we learned
Advanced integration techniques using the updated Google GenAI Python SDK.
Effective strategies for combining vision-language models with practical municipal decision-making workflows.
What's next for SmartWaste AI
Integrating real IoT weight sensors alongside computer vision for hybrid verification.
Expanding the routing engine using Google Maps Distance Matrix API for live traffic updates.
IP adress(34.50.177.3) you can enter the website with this
Log in or sign up for Devpost to join the conversation.