Monitoring Myanmar's Water Networks: Landsat-8 Imagery and Binary Water Masks

Myanmar is the geographic focus of this dataset, which contains 2,372 paired satellite image chips and binary water body masks. Each sample consists of an 8-band Landsat-8 image and a corresponding single-channel label, both at a spatial resolution of 30 meters per pixel. The data is intended for training and evaluating deep learning models for water body segmentation and environmental monitoring.

Use Cases

• Training semantic segmentation models for water body detection based on 8-band Landsat-8 imagery.
• Benchmarking adaptive deep learning frameworks for analyzing complex river networks in satellite data.
• Environmental monitoring and change detection using time-series compatible, georeferenced image chips.
• Validating remote sensing water extraction algorithms against hand-labeled binary masks.

Strengths

• Contains 2,372 precisely aligned image-label pairs, providing a substantial volume for model training.
• Labels use a clear binary classification scheme (0 for non-water, 1 for water), simplifying the segmentation task.
• Data is derived from Landsat-8 satellite imagery with a consistent 30-meter spatial resolution.
• Released under a permissive CC-BY-4.0 license, facilitating open research and reuse.

Limitations

• Column names and detailed metadata (e.g., acquisition dates, specific geographic coordinates) are not provided, limiting reproducibility.
• The stated 'Last updated' date of 2026-03-18 is in the future, casting doubt on the accuracy of this metadata field.
• The dataset is packaged in a RAR archive, which may require additional software for extraction compared to more common formats.

Provenance

Source

Landsat-8 satellite imagery, processed and labeled by author THANT ZIN TUN.

Collection Method

Remote sensing imagery processed into 128x128 pixel chips with corresponding hand-annotated binary masks.

Freshness

2026-03-18 10:04:54 (Note: This date is in the future and may be a metadata error)

Geography

Myanmar