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

3D Scene Reconstruction from a Single Viewport

Denninger, Maximilian and Triebel, Rudolph (2020) 3D Scene Reconstruction from a Single Viewport. In: 16th European Conference on Computer Vision, ECCV 2020, 16, pp. 51-67. Springer, Cham. European Conference on Computer Vision ECCV 2020, 23.-28. August 2020, Virtuell. doi: 10.1007/978-3-030-58542-6_4. ISBN 978-3-030-58541-9.

[img] PDF
9MB

Abstract

We present a novel approach to infer volumetric reconstructions from a single viewport, based only on an RGB image and a reconstructed normal image. To overcome the problem of reconstructing regions in 3D that are occluded in the 2D image, we propose to learn this information from synthetically generated high-resolution data. To do this, we introduce a deep network architecture that is specifically designed for volumetric TSDF data by featuring a specific tree net architecture. Our framework can handle a 3D resolution of 512³ by introducing a dedicated compression technique based on a modified autoencoder. Furthermore, we introduce a novel loss shaping technique for 3D data that guides the learning process towards regions where free and occupied space are close to each other. As we show in experiments on synthetic and realistic benchmark data, this leads to very good reconstruction results, both visually and in terms of quantitative measures.

Item URL in elib:https://elib.dlr.de/139323/
Document Type:Conference or Workshop Item (Speech)
Title:3D Scene Reconstruction from a Single Viewport
Authors:
AuthorsInstitution or Email of AuthorsAuthor's ORCID iD
Denninger, MaximilianMaximilian.Denninger (at) dlr.dehttps://orcid.org/0000-0002-1557-2234
Triebel, RudolphRudolph.Triebel (at) dlr.dehttps://orcid.org/0000-0002-7975-036X
Date:23 August 2020
Journal or Publication Title:16th European Conference on Computer Vision, ECCV 2020
Refereed publication:Yes
Open Access:Yes
Gold Open Access:No
In SCOPUS:No
In ISI Web of Science:No
Volume:16
DOI :10.1007/978-3-030-58542-6_4
Page Range:pp. 51-67
Publisher:Springer, Cham
Series Name:European Conference on Computer Vision
ISBN:978-3-030-58541-9
Status:Published
Keywords:Scene Reconstruction, 3D from Single Images, Space Compression, Deep Learning, Machine Learning, Neural Networks
Event Title:European Conference on Computer Vision ECCV 2020
Event Location:Virtuell
Event Type:international Conference
Event Dates:23.-28. August 2020
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Space
HGF - Program Themes:Space System Technology
DLR - Research area:Raumfahrt
DLR - Program:R SY - Space System Technology
DLR - Research theme (Project):R - Vorhaben Multisensorielle Weltmodellierung (old)
Location: Oberpfaffenhofen
Institutes and Institutions:Institute of Robotics and Mechatronics (since 2013) > Perception and Cognition
Deposited By: Denninger, Maximilian
Deposited On:08 Dec 2020 14:51
Last Modified:08 Dec 2020 14:51

Repository Staff Only: item control page

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