Case study: satellite-based street lighting assessment for urban analytics

Context

Need to assess territory lighting levels at scale without field measurements.

Task

Develop a reproducible satellite image analysis pipeline using ML.

Implementation

• Image processing and lighting feature extraction
• Model training and validation
• Geospatial result visualization

Results

• Automated analysis of territories exceeding 1,200 km²
• Significantly reduced manual assessment workload
• Created a scalable tool for infrastructure decision-making