Deep learning for anchor detection in multi-scale maps - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Deep learning for anchor detection in multi-scale maps

Résumé

Landmarks, in physical space, are salient elements in the environment that allow people to orientate themselves and to find their way in the landscape. The concept of landmark has been extended to non-physical spaces as well (Sorrows and Hirtle, 1999). In a map, a landmark can be any noticeable object or distinct pattern such as a building, a river, or even a space where there is an absence of features. When grouped, they allow people to make sense of maps (Figure 1). Automatic landmark detection in maps is a well-studied field (Figure 2), where applications range from map generalization to GPS guidance. It combines numerous research fields from machine learning, pattern recognition, statistics and data visualization (Elias, 2003). Multi-scale anchors are a specific kind of landmark in multi-scale maps that are salient in different maps at consecutive scales (Touya et al., 2020). The presence of multi-scale anchors allows a human to maintain a sense of location when zooming in and out of an interactive map. Our research explores how we can detect potential multi-scale anchors in order to make them more distinct and memorable in maps with different levels of detail, and thus make the zooming process more fluid and intuitive.
Fichier principal
Vignette du fichier
ica-abs-5-82-2022.pdf (1 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-03788316 , version 1 (26-09-2022)

Licence

Paternité

Identifiants

Citer

Quentin Potié, Guillaume Touya, Chaima Beladraoui, Amina El-Moutaouakkil, William A Mackaness. Deep learning for anchor detection in multi-scale maps. EuroCarto 2022, Sep 2022, Vienna, Austria. pp.82, ⟨10.5194/ica-abs-5-82-2022⟩. ⟨hal-03788316⟩
37 Consultations
34 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More