Libya: Road Surface Data

About this dataset

This dataset provides detailed information on road surfaces from OpenStreetMap (OSM), distinguishing between paved and unpaved roads across the region. Surface classification is derived using a hybrid deep learning approach. For methodological details, see: http://arxiv.org/abs/2410.19874 Recently, we also published a new country-level dataset based on PlanetScope satellite imagery with near-complete coverage of all arterial roads. It is available on HDX under the name Libya Planet Road Surface. The corresponding paper is available at: https://arxiv.org/abs/2512.04092 Road Surface Statistics Approximately 0.2478 million km of roads are mapped in OSM in this region. Based on AI-derived estimates: - Paved roads: 0.0573 million km (23.1418%) - Unpaved roads: 0.057 million km (22.9852%) About 0.1335 million km (53.873%) of road surface information is missing in OSM. To reduce this gap, the Mapillary-derived road surface dataset provides an additional 0.0 million km (0.0209%) of the missing surface information. Intended Use This dataset supports transportation planning, infrastructure analysis, climate emissions assessment, and GIS applications. AI Features - pred_class: Model-predicted surface class ("paved" or "unpaved") - pred_label: Binary label (0 = paved, 1 = unpaved) - osm_surface_class: Surface classification derived from OSM tags - combined_surface_osm_priority: Combined classification prioritizing OSM - combined_surface_DL_priority: Combined classification prioritizing DL prediction - n_of_predictions_used: Number of predictions used for estimation - predicted_length: Predicted length based on DL estimation (meters) - DL_mean_timestamp: Mean timestamp of predictions used OSM Attributes - name - name:en - **name:* (local language) - highway - surface - smoothness - width - lanes - oneway - bridge - layer - source Urban Classification Attributes - continent - country_iso_a2 - urban (0 = rural, 1 = urban) - urban_area - osm_id - osm_type** Data Sources The dataset originates from OpenStreetMap (OSM) and is augmented with deep learning predictions based on Mapillary street-level imagery, combined with the GHSU Global Human Settlement Urban Layer 2019 and AFRICAPOLIS 2020 urban layer. More HeiGIT datasets: https://data.humdata.org/organization/heidelberg-institute-for-geoinformation-technology More information: https://heigit.org/ Contact: communications@heigit.org