DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Multi-Task cGAN for Simultaneous Spaceborne DSM Refinement and Roof-Type Classification

Bittner, Ksenia and Körner, Marco and Fraundorfer, Friedrich and Reinartz, Peter (2019) Multi-Task cGAN for Simultaneous Spaceborne DSM Refinement and Roof-Type Classification. Remote Sensing, 11 (11), pp. 1262-1284. Multidisciplinary Digital Publishing Institute (MDPI). DOI: 10.3390/rs11111262 ISSN 2072-4292

[img] PDF - Published version

Official URL: https://www.mdpi.com/2072-4292/11/11/1262


Various deep learning applications benefit from multi-task learning with multiple regression and classification objectives by taking advantage of the similarities between individual tasks. This can result in improved learning efficiency and prediction accuracy for the task-specific models compared to separately trained models. In this paper, we make an observation of such influences for important remote sensing applications like elevation model generation and semantic segmentation tasks from the stereo half-meter resolution satellite digital surface models (DSMs). Mainly, we aim to generate good-quality DSMs with complete, as well as accurate level of detail (LoD)2-like building forms and to assign an object class label to each pixel in the DSMs. For the label assignment task, we select the roof type classification problem to distinguish between flat, non-flat, and background pixels. To realize those tasks, we train a conditional generative adversarial network (cGAN) with an objective function based on least squares residuals and an auxiliary term based on normal vectors for further roof surface refinement. Besides, we investigate recently published deep learning architectures for both tasks and develop the final end-to-end network, which combines different models, as using them first separately, they provide the best results for their individual tasks.

Item URL in elib:https://elib.dlr.de/127861/
Document Type:Article
Title:Multi-Task cGAN for Simultaneous Spaceborne DSM Refinement and Roof-Type Classification
AuthorsInstitution or Email of AuthorsAuthors ORCID iD
Bittner, KseniaKsenia.Bittner (at) dlr.dehttps://orcid.org/0000-0002-4048-3583
Körner, Marcomarco.koerner (at) tum.dehttps://orcid.org/0000-0002-9186-4175
Fraundorfer, Friedrichfraundorfer (at) icg.tugraz.athttps://orcid.org/0000-0002-5805-8892
Reinartz, Peterpeter.reinartz (at) dlr.dehttps://orcid.org/0000-0002-8122-1475
Date:28 May 2019
Journal or Publication Title:Remote Sensing
Refereed publication:Yes
Open Access:Yes
Gold Open Access:Yes
In ISI Web of Science:Yes
DOI :10.3390/rs11111262
Page Range:pp. 1262-1284
Publisher:Multidisciplinary Digital Publishing Institute (MDPI)
Keywords:multi-task learning; conditional generative adversarial networks; digital surface model; 3D scene refinement; semantic segmentation; roof type classification; urban region; satellite imagery
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Transport
HGF - Program Themes:Road Transport
DLR - Research area:Transport
DLR - Program:V ST Straßenverkehr
DLR - Research theme (Project):V - D.MoVe
Location: Oberpfaffenhofen
Institutes and Institutions:Remote Sensing Technology Institute > Photogrammetry and Image Analysis
Deposited By: Bittner, Ksenia
Deposited On:14 Jun 2019 10:26
Last Modified:14 Dec 2019 04:27

Repository Staff Only: item control page

Help & Contact
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.