2021-10-01, 08:30–09:00, Aconcagua
Built on the shoulders of the Orfeo ToolBox and TensorFlow, our software uses deep learning to remove clouds in optical images, using joint SAR/Optical Sentinel images. It is open-source, and comes with pre-trained models.
Clouds represent the main issue affecting optical satellite images. Cloud-free scenes available at specific date is crucial in a wide range of monitoring applications. Differently, Synthetic Aperture Radar (SAR) sensors provide orthogonal information with respect to optical satellite, that enable the retrieval of information lost in optical images due to cloud cover. In the context of an increasing availability of both optical and SAR images, thank to the Sentinel constellation, a number of deep learning method have emerged in recent papers. These methods aim to reconstruct optical data polluted by cloud phenomena, exploiting SAR and optical images. We present an open-source software based on the Orfeo ToolBox and TensorFlow, that provide a framework to apply methods processing Sentinel-1 and Sentinel-2 images. Our software comes with a few pre-trained models that can be used out-of-the-box to remove clouds in Sentinel-2 images from Sentinel-1 and Sentinel-2 time series.
Rémi Cresson (1)
Benjamin Commandré (1)
Nicolas Narçon (1)
Raffaele Gaetano (2)
Aurore DUPUIS (3)
Yannick TANGUY (3)
Stéphane MAY (3)
Xavier RAVE (3)
1: INRAE (French national research institute for agriculture, food and the environment
2: CIRAD (French agricultural research centre for international development)
3: CNES (French national space agency)
Education & researchTopic –
New trends: IoT, Indoor mapping, drones - UAV (unmanned aerial vehicle), Artificial intelligence - machine learning, deep learning-, geospatial data structures, real time raster analysisLevel –
2 - Basic. General basic knowledge is required.Language of the Presentation –
He is with the French National Research Institute for Agriculture, Food and Environment. His research and engineering fields include remote sensing images processing at scale, high performance computing, machine learning, and geospatial data interoperability. He is member of the Orfeo ToolBox Project Steering Committee and charter member of the OSGeo.