Navigation überspringen
Universitätsbibliothek Heidelberg
Status: Bibliographieeintrag

Verfügbarkeit
Standort: ---
Exemplare: ---
heiBIB
 Online-Ressource
Verfasst von:Chen, Jiaoyan [VerfasserIn]   i
 Zipf, Alexander [VerfasserIn]   i
Titel:Deep learning from multiple crowds
Titelzusatz:a case study of humanitarian mapping
Verf.angabe:Jiaoyan Chen, Yan Zhou, Alexander Zipf, and Hongchao Fan
Jahr:2019
Jahr des Originals:2018
Umfang:10 S.
Fussnoten:Date of Publication: 03 October 2018 ; Gesehen am 31.05.2019
Titel Quelle:Enthalten in: Institute of Electrical and Electronics EngineersIEEE transactions on geoscience and remote sensing
Ort Quelle:New York, NY : IEEE, 1964
Jahr Quelle:2019
Band/Heft Quelle:57(2019), 3, Seite 1713-1722
Abstract:Satellite images are widely applied in humanitarian mapping that labels buildings, roads, and so on for humanitarian aid and economic development. However, the labeling now is mostly done by volunteers. In this paper, we utilize deep learning to solve humanitarian mapping tasks of a mobile software named MapSwipe. The current deep learning techniques, e.g., convolutional neural network (CNN), can recognize ground objects from satellite images but rely on numerous labels for training for each specific task. We solve this problem by fusing multiple freely accessible crowdsourced geographic data and propose an active learning-based CNN training framework named MC-CNN to deal with the quality issues of the labels extracted from these data, including incompleteness (e.g., some kinds of object are not labeled) and heterogeneity (e.g., different spatial granularities). The method is evaluated with building mapping in South Malawi and road mapping in Guinea with level-18 satellite images provided by Bing Map and volunteered geographic information from OpenStreetMap, MapSwipe, and OsmAnd. The results based on multiple metrics, including Precision, Recall, F1 Score, and area under the receiver operating characteristic curve, show that MC-CNN can fuse the crowdsourced labels for higher prediction performance and be successfully applied in MapSwipe for humanitarian mapping with 85% labor saved and an overall accuracy of 0.86 achieved.
DOI:doi:10.1109/TGRS.2018.2868748
URL:Bitte beachten Sie: Dies ist ein Bibliographieeintrag. Ein Volltextzugriff für Mitglieder der Universität besteht hier nur, falls für die entsprechende Zeitschrift/den entsprechenden Sammelband ein Abonnement besteht oder es sich um einen OpenAccess-Titel handelt.

DOI: https://doi.org/10.1109/TGRS.2018.2868748
Datenträger:Online-Ressource
Sprache:eng
Sach-SW:Active learning
 active learning-based CNN training framework
 Bing Map
 building mapping
 Buildings
 convolutional neural nets
 convolutional neural network
 crowdsourced labels
 crowdsourcing
 deep learning
 economic development
 geographic information systems
 ground objects
 humanitarian aid
 humanitarian mapping
 humanitarian mapping tasks
 Labeling
 learning (artificial intelligence)
 level-18 satellite images
 Machine learning
 MapSwipe
 MC-CNN
 mobile computing
 mobile software
 multiple crowds
 multiple freely accessible crowdsourced geographic data
 multiple metrics
 numerous labels
 OpenStreetMap
 OsmAnd
 road mapping
 Roads
 satellite image
 Satellites
 Task analysis
 Training
 volunteered geographic information (VGI)
K10plus-PPN:1666566497
Verknüpfungen:→ Zeitschrift

Permanenter Link auf diesen Titel (bookmarkfähig):  https://katalog.ub.uni-heidelberg.de/titel/68394323   QR-Code
zum Seitenanfang