Attentive spatial temporal graph CNN for land cover mapping from multi temporal remote sensing data
2021
Censi, Alessandro Michele | Ienco, Dino | Gbodjo, Yawogan Jean Eudes | Pensa, Ruggero Gaetano | Interdonato, Roberto | Gaetano, Raffaele | Territoires, Environnement, Télédétection et Information Spatiale (UMR TETIS) ; Centre de Coopération Internationale en Recherche Agronomique pour le Développement (Cirad)-AgroParisTech-Centre National de la Recherche Scientifique (CNRS)-Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement (INRAE) | Università degli studi di Torino = University of Turin (UNITO) | Département Environnements et Sociétés (Cirad-ES) ; Centre de Coopération Internationale en Recherche Agronomique pour le Développement (Cirad) | French Ministry of agriculture Agricultural and Rural Development | PARCELLE Project - French Space Agency under Grant DAR CNES | ANR-16-CONV-0004,DIGITAG,Institut Convergences en Agriculture Numérique(2016) | ANR-10-EQPX-0020,GEOSUD,GEOSUD : Infrastructure nationale d'imagerie satellitaire pour la recherche sur l'environnement et les territoires et ses applications à la gestion et aux politiques publiques(2010)
International audience
Mostrar más [+] Menos [-]Inglés. Satellite image time series (SITS) collected by modern Earth Observation (EO) systems represent a valuable source of information that supports several tasks related to the monitoring of the Earth surface dynamics over large areas. A main challenge is then to design methods able to leverage the complementarity between the temporal dynamics and the spatial patterns that characterize these data structures. Focusing on land cover classification (or mapping) tasks, the majority of approaches dealing with SITS data only considers the temporal dimension, while the integration of the spatial context is frequently neglected. In this work, we propose an attentive spatial temporal graph convolutional neural network that exploits both spatial and temporal dimensions in SITS. Despite the fact that this neural network model is well suited to deal with spatio-temporal information, this is the first work that considers it for the analysis of SITS data. Experiments are conducted on two study areas characterized by different land cover landscapes and real-world operational constraints (i.e., limited labeled data due to acquisition costs). The results show that our model consistently outperforms all the competing methods obtaining a performance gain, in terms of F-Measure, of at least 5 points with respect to the best competing approaches on both benchmarks.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por AgroParisTech