A new approach to analyzing changes in multi-temporal satellite images
Great importance in remote sensing, monitoring environmental changes and land use
Björn Þór Jónsson, associate professor at Reykjavik University and Maria Eirini Pegia a PhD student at the Department of Computer Science at Reykjavik University, received the best paper award at the International Conference on Content-Based Multimedia Indexing CBMI 2022.
The award-winning paper, BiasUNet: Learning Change Detection over Sentinel-2 Image Pairs, was co-authored by Maria Eirini Pegia, Anastasia Moumtzidou, Ilias Gialampoukidis, Björn Þór Jónsson, Stefanos Vrochidis, and Ioannis Kompatsiaris. It proposes an approach to analyzing changes in multi-temporal satellite images, which are of great importance in remote sensing, monitoring environmental changes and land use. The method proposed in the award-winning paper outperforms four state-of-the-art deep learning networks on the Sentinel-2 ONERA Satellite Change Detection (OSCD) benchmark dataset, both in terms of precision and quality.
It is always an honour to receive such recognition for your research and a large part of our work as scientists is to participate in conferences like this. As part of the recognition, we are now invited to work on a journal article where we approach the method from various angles and continue to further develop our study.
Björn Þór has extensive experience in working with large multimedia collections and Maria works in a laboratory in her native Greece that works extensively with satellite data. She has a background in mathematics, and in the CBMI paper, she uses a mathematical approach based on continually reassessing probability as more data is examined, giving the opportunity to reach a good model conclusion faster.
