Latvia Road Surface Classification from OpenStreetMap and AI Predictions

HeiGIT's dataset provides AI-derived road surface classifications for Latvia, distinguishing paved and unpaved roads. It covers approximately 127,900 km of roads mapped in OpenStreetMap, with 19.5% paved and 32.5% unpaved according to model predictions. The data was last updated on March 2, 2026, and combines OSM attributes with deep learning predictions from Mapillary imagery.

Use Cases

• Transportation planning based on road surface type and length statistics.
• Infrastructure analysis based on attributes like highway type, surface, and smoothness.
• Climate emissions assessment based on the proportion of paved versus unpaved roads.
• GIS applications for routing and accessibility based on urban/rural classification.

Strengths

• Includes AI-derived predictions for road surface type, filling gaps in OSM data.
• Combines multiple data sources: OSM, Mapillary imagery, and urban classification layers.
• Provides detailed statistics, such as 48% of road surface information being missing in OSM.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

Heidelberg Institute for Geoinformation Technology (HeiGIT)

Collection Method

Derived from OpenStreetMap and augmented with deep learning predictions based on Mapillary street-level imagery.

Time Range

The mean timestamp of predictions is included, but the overall temporal coverage is not specified.

Freshness

Last updated 2026-03-02 17:28:23.338525; freshness should be verified.

Geography

Latvia