AI4Boundaries: an open AI-ready dataset to map field boundaries with Sentinel-2 and aerial photography

Field boundaries are at the core of many agricultural applications and are a key enabler for operational monitoring of agricultural production to support food security. Recent scientific progress in deep learning methods has highlighted the capacity to extract field boundaries from satellite and aerial images. So far, no standard data set exists to easily and robustly benchmark models and progress the state of the art. The absence of such benchmark data further impedes proper comparison against existing methods. As a result, it is currently impossible to compare and benchmark new and existing methods. To fill these gaps, we introduce AI4Boundaries, a data set of images and labels readily usable to train and compare models on the task of field boundary detection: (i) a 10-m Sentinel-2 monthly composites, (ii) a 1-m orthophoto data set. All labels have been sourced from GSAA data that have been made openly available (Austria, Catalonia, France, Luxembourg, the Netherlands, Slovenia, and Sweden). Data were selected following a stratified random sampling drawn based on two landscape fragmentation metrics. Both datasets are provided with the corresponding ground-truth parcel delineation (2.5 M parcels covering 47,105 km2). Besides providing this open dataset to foster computer vision developments of parcel delineation methods, we discuss perspectives and limitations of the dataset for various types of applications in the agriculture domain and consider possible further improvements.

D'ANDRIMONT Raphael;  CLAVERIE Martin;  KEMPENEERS Pieter;  MURARO Davide;  YORDANOV Momchil;  PERESSUTTI Devis;  BATIC Matej;  WALDNER Francois; 

2023-01-23

COPERNICUS GESELLSCHAFT MBH

JRC130529

1866-3508 (online),   

https://essd.copernicus.org/articles/15/317/2023/,    https://publications.jrc.ec.europa.eu/repository/handle/JRC130529,   

10.5194/essd-15-317-2023 (online),