Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 1
Encontrada(s) 1 página(s)

Dipòsit Digital de Documents de la UAB
Publicación Artículo preliminar (WorkingPaper)

Monitoring war destruction from space using machine learning

  • Gröger, Andre|||0000-0002-1482-9392
  • Mueller, Hannes|||0000-0001-7667-3971
  • Hersh, Jonathan|||0000-0001-6786-5162
  • Matranga, Andrea|||0000-0003-3724-9937
  • Serrat, Joan|||0000-0002-4554-199X
Satellite imagery is becoming ubiquitous and is released with ever higher frequency. Research hasdemonstrated that Artificial Intelligence (AI) applied to satellite imagery holds promise for automateddetection of war-related building destruction. While these results are promising, monitoring inreal-world applications requires consistently high precision, especially when destruction is sparseand detecting destroyed buildings is equivalent to looking for a needle in a haystack. We demonstratethat exploiting the persistent nature of building destruction can substantially improve the trainingof automated destruction monitoring. We also propose an additional machine learning stage thatleverages images of surrounding areas and multiple successive images of the same area which furtherimproves detection significantly. By combining these steps, we construct an automated classificationof building destruction which allows real-world applications and we illustrate this in the context ofthe Syrian civil war.