FLAIR : a Country-Scale Land Cover Semantic Segmentation Dataset From Multi-Source Optical Imagery

We introduce the French Land cover from Aerospace ImageRy (FLAIR), an extensive dataset from the French National Institute of Geographical and Forest Information (IGN) that provides a unique and rich resource for large-scale geospatial analysis. FLAIR contains high-resolution aerial imagery with a ground sample distance of 20 cm and over 20 billion individually labeled pixels for precise landcover classification. The dataset also integrates temporal and spectral data from optical satellite time series. FLAIR thus combines data with varying spatial, spectral, and temporal resolutions across over 817 km 2 of acquisitions representing the full landscape diversity of France. This diversity makes FLAIR a valuable resource for the development and evaluation of novel methods for large-scale land-cover semantic segmentation and raises significant challenges in terms of computer vision, data fusion, and geospatial analysis. We also provide powerful uni-and multi-sensor baseline models that can be employed to assess algorithm's performance and for downstream applications. Through its extent and the quality of its annotation, FLAIR aims to spur improvements in monitoring and understanding key anthropogenic development indicators such as urban growth, deforestation, and soil artificialization. Dataset and codes can be accessed at https://ignf.github.io/FLAIR/ Context According to a 2015 report by the Food and Agriculture Organization of the United Nations (FAO) [1] , approximately 75% of the world's soils are in fair, poor, or very poor condition. This degradation poses significant threats to the health and long-term sustainability of ecosystems. Healthy soils provide invaluable ecosystem services, including: (i) providing natural habitats for numerous plant and animal species [2], (ii) acting as the largest carbon sink, surpassing the atmosphere and all combined biomass [3], and (iii) functioning as a rainwater reservoir, supporting food production and storing freshwater [4] . The degradation of soils and biodiversity is largely attributed to land artificialization [1], which causes long-term damage to the biological, hydrological, climatic, and agronomic functions of the soil due to its occupation or use [5, 6] . In order to effectively monitor and manage land artificialization, public authorities have expressed the need for scalable land-cover monitoring tools. With the increasing availability of high-quality Earth Observation (EO) data, the French National Institute 37th Conference on Neural Information Processing Systems (NeurIPS 2023) Track on Datasets and Benchmarks.