Satellite Imagery Land Use Classification Dataset

This dataset contains satellite images annotated with land use categories for training computer vision models. It includes 50,000 high-resolution scenes across 10 land cover classes such as urban, agricultural, forest, water, and barren land. Images are captured by Sentinel-2 between 2018 and 2022 across North America and Europe.

Use Cases

• Train a semantic segmentation model on satellite imagery to classify land use types (inferred from domain, verify after download)

Strengths

• 50,000 annotated images
• 10 balanced land cover classes
• High-resolution Sentinel-2 imagery
• Geographic diversity across two continents

Limitations

• Metadata on acquisition parameters is limited
• Last update date is unknown
• Cloud cover may vary per scene

Provenance

Source

European Space Agency (ESA) Copernicus program

Collection Method

Collected from Sentinel-2 satellite open data portal

Time Range

2018-2022

Geography

North America and Europe