Serbia Road Surface and Passability for 14,400 km of Arterial Roads

Covering 14,400 km of arterial roads in Serbia, this dataset provides AI-derived surface, width, and passability metrics generated by HeiGIT from 2020 and 2024 PlanetScope imagery. It integrates OpenStreetMap attributes with deep-learning predictions to fill data gaps for approximately 19% of the network lacking surface tags.

Use Cases

• Analyzing infrastructure development by tracking transitions in the 'Road Surface Change' field
• Logistics planning for heavy vehicles using 'Road Width Class' to identify segments with high blockage risk
• Humanitarian access modeling by filtering road segments based on the 'Humanitarian Passability Index (HPI)' score

Strengths

• 89.2% accuracy for AI-derived surface models
• Covers 14,400 km of the Serbian arterial road network
• Fills surface data gaps for 2,800 km of roads missing tags in OpenStreetMap

Limitations

• Excludes minor road networks such as residential and local streets
• AI-derived predictions may contain errors inherent to satellite imagery analysis compared to ground-level surveys

Provenance

Source

HeiGIT (Heidelberg Institute for Geoinformation Technology)

Collection Method

AI-derived predictions from PlanetScope satellite imagery (2020 and 2024) integrated with OpenStreetMap road segments.

Time Range

2020-2024

Freshness

Last updated March 2026; contains imagery-derived data from 2020 and 2024.

Geography

Serbia