<div class="csl-bib-body">
<div class="csl-entry">Jagschitz, P. (2021). <i>Introducing depth information to 2D segmentation output of neural network</i> [Diploma Thesis, Technische Universität Wien]. reposiTUm. https://doi.org/10.34726/hss.2021.61287</div>
</div>
-
dc.identifier.uri
https://doi.org/10.34726/hss.2021.61287
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/18527
-
dc.description
Abweichender Titel nach Übersetzung der Verfasserin/des Verfassers
-
dc.description.abstract
This thesis evaluates the effects depth information has on estimation outputs of neural networks by including the depth data as post processing step. The algorithm implemented looks for planes in the depth point cloud via RANSAC corresponding to the areas determined as floor or wall in the estimation results of the neural network. The RANSAC results are then used to correct the segmentation by the means of nearest neighbour search. The results are analysed on different threshold level, which define the smallest objects to be considered during analysis. As neural network the AdapNet++ [1], a 50 layerresidual NN, is used. Whilst the NN is trained on the Scannet dataset, the implementation of depth data is performed on the NYUv2 dataset. The results show an average increase of segmentation accuracy of 1.54% over the different thresholds, peaking at 2.73% for objects bigger than 10000 pixels. The overall prediction accuracy on pixel level decreases after the introduction of depthdata by -0.22%.
en
dc.language
English
-
dc.language.iso
en
-
dc.rights.uri
http://rightsstatements.org/vocab/InC/1.0/
-
dc.subject
Objekterkennung
de
dc.subject
Mobiler Roboter
de
dc.subject
Object recognition
en
dc.subject
mobile robot
en
dc.title
Introducing depth information to 2D segmentation output of neural network
en
dc.title.alternative
Objekterkennung mit semantischer Information am mobilen Roboter
de
dc.type
Thesis
en
dc.type
Hochschulschrift
de
dc.rights.license
In Copyright
en
dc.rights.license
Urheberrechtsschutz
de
dc.identifier.doi
10.34726/hss.2021.61287
-
dc.contributor.affiliation
TU Wien, Österreich
-
dc.rights.holder
Patrick Jagschitz
-
dc.publisher.place
Wien
-
tuw.version
vor
-
tuw.thesisinformation
Technische Universität Wien
-
dc.contributor.assistant
Patten, Timothy Michael
-
tuw.publication.orgunit
E376 - Institut für Automatisierungs- und Regelungstechnik