UAV-Based Road Defect Detection Using Hybrid Machine Learning Approach with Web GIS Visualization

Urban roads in fast-growing cities fall apart quickly, and everyone feels the impact—traffic slows down, accidents happen, and the city’s economy takes a hit. The old way of checking roads—sending people out to inspect them on foot—just doesn’t cut it anymore. It’s slow, expensive, and puts workers in harm’s way. So, we’ve built something better: an automated system that uses drones and AI to keep an eye on road conditions.

Here’s how it works. Drones fly over city streets, snapping high-resolution images that pick up everything from big potholes to tiny cracks. We run these images through our analytics pipeline. First, we use classic machine learning to weed out the stretches of road that are still in good shape. That way, the system doesn’t waste time on areas that don’t need attention.

Next, we use a deep learning model—based on YOLO, which stands for “You Only Look Once”—to hunt down and label the actual problem spots. We’ve trained this model using annotated drone photos, so it can handle tricky lighting or weird road surfaces. The model doesn’t just spot the defects—it also nails down where they are, how big they’ve gotten, and how bad the damage is.

But spotting problems isn’t enough. City agencies need to see this info and act on it, fast. So, we’ve built a web portal using OpenLayers and PostGIS that maps out every defect. Maintenance crews can sort issues by type or severity, pull up interactive maps, and even generate reports to plan repairs.

This whole setup is practical, affordable, and scales up easily for any city that wants to take road maintenance seriously. By bringing together drones, AI, and smart mapping, we’re giving city managers the real-time, reliable data they need to keep roads safe and traffic moving. And honestly, this system can help any city make smarter decisions about their roads and urban development.