Cross-year multi-modal image retrieval using siamese networks
Résumé
This paper introduces a multi-modal network that learns to retrieve by content vertical aerial images of French urban and rural territories taken about 15 years apart. This means it should be invariant against a big range of changes as the (nat-ural) landscape evolves over time. It leverages the original images and semantically segmented and labeled regions. The core of the method is a Siamese network that learns to extract features from corresponding image pairs across time. These descriptors are discriminative enough, such that a simple kNN classifier on top, suffices as final geo-matching criteria. The method outperformed SOTA "off-the-shelf" image descrip-tors GEM and ResNet50 on the new aerial images dataset.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |
![]()
Référence hal-04421094 Autre Margarita Khokhlova, Nathalie Abadie, Valérie Gouet-Brunet, Liming Chen. [Data] Geographical Entities With Neighborhood Relationships. 2020, http://alegoria.ign.fr/en/GENR_dataset. ⟨hal-04421094⟩