Reference field boundaries (paper: FieldSeg: A scalable agricultural field extraction framework based on the Segment Anything Model and 10-m Sentinel-2 imagery)
Domain:
agriculturegeospatialRecord type:
dataset
Creator:
Publisher:
Host:
Reference field boundaries dataset generated in the paper"FieldSeg: A scalable agric…".
A hand-annotated field boundary dataset (2022) covering 8 10x10 km areas across the world is made available. The study areas are located in Argentina, Australia, Brazil, China, South Africa, Spain, USA-California, and USA-Iowa.
This dataset contains two files:
- reference_field_boundaries.gpkg: hand-annotated dataset, with polygons defining the field boundaries.
- study_areas.gpkg: polygons defining the limits of the study areas and additional metadata about each area.
More information on how this dataset was prepared is available in the paper "FieldSeg: A scalable agric….
Similar
Segment Anything Model and Fully Convolutional Data Description for Plant Multi-Disease Detection on Field ImagesHigh resolution remote sensing imagery and derived agricultural field boundaries in EthiopiaFields of The World: A Field Guide for Extracting Agricultural Field Boundaries"Dataset for the paper: A First Systematic Assessment of the Segment Anything Model for Zero-shot SAR-based Flood Mapping"Surface moisture and irrigation mapping at agricultural field scale using the synergy Sentinel-1/Sentinel-2 dataMapping Field-level Maize Yields in Ethiopian Smallholder Systems Using Sentinel-2 Imagery