Experiments in the automatic segmentation of anchors using deep learning techniques - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Experiments in the automatic segmentation of anchors using deep learning techniques

Résumé

The way we navigate in maps is similar to the way we navigate in real life: we look for landmarks to remember, and we create our own mind map. We define anchors in multi-scale maps as landmarks that are memorable through multiple consecutive scales. The ultimate goal of this work is to automatically detect anchors in existing maps at any zoom level. This would allow us to detect ruptures in anchor continuity and thus learn where the map could be improved to allow more seamless transitions between scales without becoming confused or lost as we zoom across the map.
Fichier principal
Vignette du fichier
01_CR_CartoAI_Potie.pdf (516.81 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-04225030 , version 1 (02-10-2023)

Licence

Identifiants

  • HAL Id : hal-04225030 , version 1

Citer

Quentin Potié, Guillaume Touya, William A Mackaness. Experiments in the automatic segmentation of anchors using deep learning techniques. CartoAI: AI for cartography, Sep 2023, Leeds, United Kingdom. ⟨hal-04225030⟩
21 Consultations
25 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More