RAMP Building Footprints for Sylhet, Bangladesh with 30cm Satellite Imagery

2022 Maxar satellite imagery at 30cm resolution is paired with building footprint vector labels for Sylhet, Bangladesh. The dataset contains 16,217 image tiles and labels for 135,375 individual buildings. It was produced for the RAMP project by MLHUB and released on NASA EarthData in 2023.

Use Cases

• Train building detection models based on high-resolution satellite imagery and corresponding vector labels.
• Develop semantic segmentation algorithms for identifying structures in peri-urban and rural landscapes.
• Analyze building density and distribution patterns in a flood-affected region of Bangladesh.
• Benchmark model performance on a dataset containing over 135,000 annotated buildings.
• Study land use in areas characterized by rivers and agriculture using paired image and geospatial data.

Strengths

• Contains 16,217 high-resolution (30cm) image and label tile pairs.
• Includes annotations for 135,375 individual building footprints.
• Imagery is sourced from a specific, recent release (Maxar ODP 2022).

Limitations

• Dataset is labeled as a Tier 2 RAMP dataset, meaning it has not been thoroughly reviewed and improved.
• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

MLHUB for the RAMP project, hosted on NASA EarthData.

Collection Method

Satellite imagery sourced from Maxar ODP 2022 release, with corresponding building footprint labels generated.

Time Range

Imagery from 2022.

Freshness

Last updated 2023-01-01 00:00:00; freshness should be verified.

Geography

Sylhet, Bangladesh.