3D Map Reconstruction from Single Satellite Image Using a Deep Monocular Depth Network

Changmin Son, Soon Yong Park

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations

Abstract

In this paper, we propose a 3D reconstruction scheme from single image with deep monocular depth estimation network, BTS (From Big to Small: Multi-Scale Local Planar Guidance for Monocular Depth Estimation) [1]. Furthermore, we expand it to height estimation focused building from remote sensing images. To address these issues, we substitute depth estimation loss function with height estimation loss function. Moreover, considering improving the quality of the building height map and looking as similar as possible to the ground-truth view, we apply building adaptive loss function.

Original languageEnglish
Title of host publicationICUFN 2022 - 13th International Conference on Ubiquitous and Future Networks
PublisherIEEE Computer Society
Pages5-7
Number of pages3
ISBN (Electronic)9781665485500
DOIs
StatePublished - 2022
Event13th International Conference on Ubiquitous and Future Networks, ICUFN 2022 - Virtual, Barcelona, Spain
Duration: 5 Jul 20228 Jul 2022

Publication series

NameInternational Conference on Ubiquitous and Future Networks, ICUFN
Volume2022-July
ISSN (Print)2165-8528
ISSN (Electronic)2165-8536

Conference

Conference13th International Conference on Ubiquitous and Future Networks, ICUFN 2022
Country/TerritorySpain
CityVirtual, Barcelona
Period5/07/228/07/22

Keywords

  • 3D map
  • Deep learning
  • Monocular height estimation
  • Remote sensing image

Fingerprint

Dive into the research topics of '3D Map Reconstruction from Single Satellite Image Using a Deep Monocular Depth Network'. Together they form a unique fingerprint.

Cite this