Browse Datasets

South Africa Crop Type

Fused Sentinel-1/2 Imagery for Crop Mapping

semantic-segmentation data-fusion computer-vision ~10GB CC BY 4.0
Access Dataset

Description

Fused Sentinel-1 (radar) and Sentinel-2 (multispectral) imagery for crop-field classification in South Africa’s Western Cape, produced for the Radiant Earth Spot the Crop Challenge. 10 m resolution rasters with field-identification and crop-type labels.

Quick Start

Python 
# Browse and download tiles from Source Cooperative / Sentinel Hub:
#   https://collections.eurodatacube.com/south-africa-crops-competition/
# Read the GeoTIFF rasters with rasterio
import rasterio

Project Ideas

Optical + Radar Fusion

Build a model fusing Sentinel-2 optical and Sentinel-1 radar bands and show how radar penetrates cloud cover.

Advanced

Temporal Segmentation

Segment crop fields using the multi-date image stack across the growing season.

Advanced

Ethical Considerations

  • Spatial data on smallholder fields should be handled to avoid exposing localized economic vulnerabilities.