Road Surface Classification for Micronesia from OpenStreetMap and AI

A 2026 dataset from the Heidelberg Institute for Geoinformation Technology (HeiGIT) provides AI-derived road surface classifications for the Federated States of Micronesia. It distinguishes paved and unpaved roads, covering approximately 0.8 km of mapped roads, with 0.2 km classified as paved and 0.2 km as unpaved. The data originates from OpenStreetMap and is augmented with deep learning predictions from Mapillary imagery.

Use Cases

• Transportation planning based on road surface type and highway classification.
• Infrastructure analysis based on attributes like surface, smoothness, width, and bridge status.
• Climate emissions assessment based on paved vs. unpaved road network composition.
• GIS applications for urban-rural classification using the urban attribute.

Strengths

• Integrates multiple data sources: OpenStreetMap, Mapillary imagery, and urban layers from GHSU and AFRICAPOLIS.
• Provides AI-derived predictions for road surface where OSM data is missing, covering an estimated 12.5671% of the missing information.
• Includes detailed OSM attributes like highway type, surface, smoothness, and width for analysis.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count and file size are unknown, which may limit suitability assessment.
• Approximately 51.0338% of road surface information was originally missing in the underlying OSM data.

Provenance

Source

Heidelberg Institute for Geoinformation Technology (HeiGIT)

Collection Method

Derived from OpenStreetMap and augmented with a hybrid deep learning approach using Mapillary street-level imagery.

Time Range

The mean timestamp of predictions is recorded in the DL_mean_timestamp field.

Freshness

Last updated 2026-03-02 17:27:30.788447; freshness should be verified.

Geography

Federated States of Micronesia