Bangladesh Arterial Road Surface and Passability Data (2020-2024)

This dataset maps approximately 19,800 km of arterial roads in Bangladesh, providing AI-derived surface types, widths, and passability scores produced by HeiGIT. Using PlanetScope satellite imagery from 2020 and 2024, the data supplements OpenStreetMap (OSM) attributes with deep-learning predictions that achieve 89.2% accuracy.

Use Cases

• Evaluating heavy vehicle accessibility using the Road Width Class and Humanitarian Passability Index (HPI)
• Identifying infrastructure improvements by filtering for 'unpaved to paved' transitions in the Road Surface Change column
• Enhancing transport models by replacing missing OSM surface tags with DL Surface Prediction 2024 values

Strengths

• 89.2% surface prediction accuracy
• Covers 19,800 km of arterial roads
• Multi-temporal comparison between 2020 and 2024
• Fills surface tag gaps for 40% of OSM road segments

Limitations

• Geographic scope restricted to Bangladesh
• Excludes minor and residential road classes
• AI-derived predictions may have residual errors compared to ground-level verification

Provenance

Source

Heidelberg Institute for Geoinformation Technology (HeiGIT)

Collection Method

Deep learning analysis of PlanetScope satellite imagery combined with OpenStreetMap road network data.

Time Range

2020-2024

Freshness

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

Geography

Bangladesh