CROSS-YEAR MULTI-MODAL IMAGE RETRIEVAL USING SIAMESE NETWORKS

This paper introduces a multi-modal network that learns to retrievebycontentverticalaerialimagesofFrenchurbanand rural territories taken about 15 years apart. This means it shouldbeinvariantagainstabigrangeofchangesasthe(natural) landscape evolves over time. It leverages the original images and semantically segmented and labeled regions. The coreofthemethodisaSiamesenetworkthatlearnstoextract features from corresponding image pairs across time. These descriptorsarediscriminativeenough,suchthatasimplekNN classifier on top, suffices as final geo-matching criteria. The method outperformed SOTA ”off-the-shelf” image descriptors GEM and ResNet50 on the new aerial images dataset.