In Roeselare, Belgium, we assessed the quality of road markings from satellite and panchromatic imagery, grading every line segment as Good, Medium or Bad across the whole city.
The challenge
The city needed an objective, repeatable read on line visibility and integrity. Manual surveys were slow, inconsistent and disruptive, with no automated way to separate crisp markings from faded or failing ones.
Our approach
We fused high-resolution panchromatic detail with multispectral satellite data and ran deep-learning models that:
- detected continuous and dashed line segments;
- judged quality from contrast, continuity and visibility against the road surface;
- graded each segment as Good, Medium (partially faded but still discernible) or Bad.
Results & benefits
- A city-wide digital map of road-line quality for Roeselare.
- Objective, automated, scalable grading that prioritises repainting where it is needed most.
