2021 · 1.2K citations · 10 references
ComputingCloud ComputingEarth SciencesGlobal Navigation Satellite SystemsMachine LearningRemote SensingDeep LearningGeospatial Machine LearningLand UseLand Use Change DetectionRemote Sensing Image AnalysisRemote Sensing ImageryGeospatial AnalysisGlobal Environmental ChangeHigh-resolution Satellite Image AnalysisHigh-resolution Satellite Imagery AnalysisRemote Sensing SegmentationSatellite Imagery AnalysisScalable Cloud ComputingSentinel Systems
Land use/land cover maps are essential for monitoring environmental change and risk, and Sentinel‑2 imagery—combined with recent deep‑learning advances—offers the high‑resolution, scalable data needed for automated, actionable geospatial analysis. A deep‑learning segmentation model trained on over 5 billion human‑labeled Sentinel‑2 pixels produces a global 10 m resolution LULC map with state‑of‑the‑art accuracy, enabling automated mapping from time‑series observations.
Land use/land cover (LULC) maps are foundational geospatial data products needed by analysts and decision makers across governments, civil society, industry, and finance to monitor global environmental change and measure risk to sustainable livelihoods and development. There is a strong need for high-level, automated geospatial analysis products that turn these pixels into actionable insights for non-geospatial experts. The Sentinel 2 satellites, first launched in mid-2015, are excellent candidates for LULC mapping due to their high spatial, spectral, and temporal resolution. Advances in deep learning and scalable cloud-based compute now provide the analysis capability required to unlock the value in global satellite imagery observations. Based on a novel, very large dataset of over 5 billion human-labeled Sentinel-2 pixels, we developed and deployed a deep learning segmentation model on Sentinel-2 data to create a global LULC map at 10m resolution that achieves state-of-the-art accuracy and enables automated LULC mapping from time series observations.
10
Global land change from 1982 to 2016
Xiao‐Peng Song, Matthew C. Hansen, Stephen V. Stehman et al. · Nature · 2018 · 1.9K citations · Full text
Natural Sciences, Land-use Planning, Environmental Change +9
Pontus Olofsson, Giles M. Foody, Stephen V. Stehman et al. · Remote Sensing of Environment · 2012 · 1K citations
Copernicus Global Land Cover Layers—Collection 2
Marcel Buchhorn, Myroslava Lesiv, Nandin‐Erdene Tsendbazar et al. · Remote Sensing · 2020 · 889 citations · Full text
Dropout Training as Adaptive Regularization
Stefan Wager, Sida Wang, Percy Liang · arXiv (Cornell University) · 2013 · 274 citations · Full text