The Challenge

Traditional methods for mapping traffic signs and public lighting poles rely on manual field surveys. This process is slow, expensive, and prone to inconsistencies, creating challenges for effective infrastructure management. The goal was to test the feasibility of an automated, AI-driven solution.

The Technical Solution

A prototype computer vision system was developed to automatically process street-level imagery and map key infrastructure.

• AI-Powered Detection: A YOLO (You Only Look Once) object detection model was trained to accurately identify and classify traffic signs and public lighting poles from video frames.
• Geospatial Mapping: The system calculates the real-world GPS coordinates for each detected object, linking the visual data to a physical location.
• Interactive Visualization: A user-friendly dashboard was built with Streamlit and Leaflet (OpenStreetMap) to display the detected objects on an interactive map, allowing for simple review and validation of the results.

Results and Impact

The prototype successfully proved that an AI-based approach could dramatically improve the infrastructure mapping process. The system demonstrated the ability to:

• Increase Efficiency: Significantly accelerate the analysis of video footage compared to manual inspection.
• Improve Accuracy: Provide a high degree of precision and consistency in object recognition.
• Deliver Actionable Data: Present results in an intuitive visual format, making the data immediately useful for planners and managers.

This project was developed as part of an EDIH Adria 'Test Before Invest' initiative. For more details on the collaboration, you can read the official success story.