Deep domain adaptation by weighted entropy minimization for the classification of aerial images

Download statistics - Document (COUNTER):

Wittich, D.: Deep domain adaptation by weighted entropy minimization for the classification of aerial images. In: Paparoditis, N. et.al. (Eds.): XXIV ISPRS Congress, Commission II : edition 2020. Katlenburg-Lindau : Copernicus Publications, 2020 (ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences ; 5,2), S. 591-598. DOI: https://doi.org/10.5194/isprs-annals-V-2-2020-591-2020

Repository version

To cite the version in the repository, please use this identifier: https://doi.org/10.15488/10881

Selected time period:

year: 
month: 

Sum total of downloads: 47




Thumbnail
Abstract: 
Fully convolutional neural networks (FCN) are successfully used for the automated pixel-wise classification of aerial images and possibly additional data. However, they require many labelled training samples to perform well. One approach addressing this issue is semi-supervised domain adaptation (SSDA). Here, labelled training samples from a source domain and unlabelled samples from a target domain are used jointly to obtain a target domain classifier, without requiring any labelled samples from the target domain. In this paper, a two-step approach for SSDA is proposed. The first step corresponds to a supervised training on the source domain, making use of strong data augmentation to increase the initial performance on the target domain. Secondly, the model is adapted by entropy minimization using a novel weighting strategy. The approach is evaluated on the basis of five domains, corresponding to five cities. Several training variants and adaptation scenarios are tested, indicating that proper data augmentation can already improve the initial target domain performance significantly resulting in an average overall accuracy of 77.5%. The weighted entropy minimization improves the overall accuracy on the target domains in 19 out of 20 scenarios on average by 1.8%. In all experiments a novel FCN architecture is used that yields results comparable to those of the best-performing models on the ISPRS labelling challenge while having an order of magnitude fewer parameters than commonly used FCNs. © 2020 Copernicus GmbH. All rights reserved.
License of this version: CC BY 4.0 Unported
Document Type: bookPart
Publishing status: publishedVersion
Issue Date: 2020
Appears in Collections:Fakultät für Bauingenieurwesen und Geodäsie

distribution of downloads over the selected time period:

downloads by country:

pos. country downloads
total perc.
1 image of flag of Germany Germany 14 29.79%
2 image of flag of United States United States 8 17.02%
3 image of flag of China China 7 14.89%
4 image of flag of Netherlands Netherlands 4 8.51%
5 image of flag of No geo information available No geo information available 2 4.26%
6 image of flag of Taiwan Taiwan 2 4.26%
7 image of flag of Hong Kong Hong Kong 2 4.26%
8 image of flag of Czech Republic Czech Republic 2 4.26%
9 image of flag of Indonesia Indonesia 1 2.13%
10 image of flag of Austria Austria 1 2.13%
    other countries 4 8.51%

Further download figures and rankings:


Hinweis

Zur Erhebung der Downloadstatistiken kommen entsprechend dem „COUNTER Code of Practice for e-Resources“ international anerkannte Regeln und Normen zur Anwendung. COUNTER ist eine internationale Non-Profit-Organisation, in der Bibliotheksverbände, Datenbankanbieter und Verlage gemeinsam an Standards zur Erhebung, Speicherung und Verarbeitung von Nutzungsdaten elektronischer Ressourcen arbeiten, welche so Objektivität und Vergleichbarkeit gewährleisten sollen. Es werden hierbei ausschließlich Zugriffe auf die entsprechenden Volltexte ausgewertet, keine Aufrufe der Website an sich.

Search the repository


Browse