Domain adaptation for change detection in heterogeneous remote sensing imagery

Details

Supervisors

De Vleeschouwer, Christophe ; Boracchi, Giacomo

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en informatique, à finalité spécialisée

Abstract

Change detection is an important and challenging problem in remote sensing and earth observation. Many applications can benefit from using heterogeneous imagery in this context. However, it is a very challenging task as the imagery need to be transformed into a shared domain before being able to be compared. Existing methodologies consider deep learning models trained on datasets of paired images to be able to learn the needed translation function. In this work, we explore the possibility to address this described domain adaptation task without having to rely on paired images during training, thus allowing to build scalable and generalizable models.