Sentinel-2 Agricultural Land Cover Labels for Crop Residue and Soil

Manually labeled pixel samples for agricultural land cover classification. The dataset includes annotations for crop residue, burn scars, and soil types, derived from Sentinel-2 satellite imagery. The author, organization, and specific geographic scope are unknown.

Use Cases

• Train semantic segmentation models for crop residue detection based on manual pixel labels.
• Classify soil types from satellite imagery using the provided labeled samples.
• Identify and map agricultural burn scars using annotated remote sensing data.
• Benchmark land cover classification algorithms on manually verified agricultural pixels.

Strengths

• Data is manually labeled, which likely indicates high-quality ground truth for training.
• Focuses on specific agricultural land cover classes: crop residue, burn scar, and soil.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

kaggle

Collection Method

Manually labeled pixel samples.