Open Access
Open access

Convolutional Occupancy Networks

Songyou Peng 1, 2
Michael Niemeyer 2, 3
Lars Mescheder 2, 4
MARC POLLEFEYS 1, 5
Andreas Geiger 2, 3
Publication typeBook Chapter
Publication date2020-12-02
scimago Q2
SJR0.352
CiteScore2.4
Impact factor
ISSN03029743, 16113349, 18612075, 18612083
Abstract
Recently, implicit neural representations have gained popularity for learning-based 3D reconstruction. While demonstrating promising results, most implicit approaches are limited to comparably simple geometry of single objects and do not scale to more complicated or large-scale scenes. The key limiting factor of implicit methods is their simple fully-connected network architecture which does not allow for integrating local information in the observations or incorporating inductive biases such as translational equivariance. In this paper, we propose Convolutional Occupancy Networks, a more flexible implicit representation for detailed reconstruction of objects and 3D scenes. By combining convolutional encoders with implicit occupancy decoders, our model incorporates inductive biases, enabling structured reasoning in 3D space. We investigate the effectiveness of the proposed representation by reconstructing complex geometry from noisy point clouds and low-resolution voxel representations. We empirically find that our method enables the fine-grained implicit 3D reconstruction of single objects, scales to large indoor scenes, and generalizes well from synthetic to real data.
Found 
Found 

Top-30

Journals

10
20
30
40
50
60
70
80
90
Lecture Notes in Computer Science
81 publications, 15.43%
IEEE Transactions on Pattern Analysis and Machine Intelligence
27 publications, 5.14%
IEEE Transactions on Visualization and Computer Graphics
21 publications, 4%
ACM Transactions on Graphics
19 publications, 3.62%
IEEE Access
9 publications, 1.71%
IEEE Robotics and Automation Letters
8 publications, 1.52%
Computers and Graphics
6 publications, 1.14%
Remote Sensing
5 publications, 0.95%
Computer Graphics Forum
4 publications, 0.76%
IEEE Transactions on Image Processing
4 publications, 0.76%
IEEE Transactions on Multimedia
4 publications, 0.76%
IEEE Transactions on Circuits and Systems for Video Technology
3 publications, 0.57%
IEEE Transactions on Robotics
3 publications, 0.57%
Graphical Models
3 publications, 0.57%
ISPRS Journal of Photogrammetry and Remote Sensing
3 publications, 0.57%
Medical Image Analysis
3 publications, 0.57%
Communications in Computer and Information Science
3 publications, 0.57%
International Journal of Computer Vision
3 publications, 0.57%
IEEE Transactions on Medical Imaging
3 publications, 0.57%
Journal of Electronic Imaging
2 publications, 0.38%
Sensors
2 publications, 0.38%
IEEE Sensors Journal
2 publications, 0.38%
Pattern Recognition
2 publications, 0.38%
ACM Computing Surveys
2 publications, 0.38%
IEEE Transactions on Geoscience and Remote Sensing
2 publications, 0.38%
IEEE Signal Processing Letters
2 publications, 0.38%
Computer Animation and Virtual Worlds
2 publications, 0.38%
Displays
2 publications, 0.38%
Mathematics
1 publication, 0.19%
10
20
30
40
50
60
70
80
90

Publishers

50
100
150
200
250
300
350
Institute of Electrical and Electronics Engineers (IEEE)
321 publications, 61.14%
Springer Nature
96 publications, 18.29%
Association for Computing Machinery (ACM)
40 publications, 7.62%
Elsevier
32 publications, 6.1%
MDPI
12 publications, 2.29%
Wiley
6 publications, 1.14%
SPIE-Intl Soc Optical Eng
4 publications, 0.76%
Taylor & Francis
2 publications, 0.38%
Tsinghua University Press
1 publication, 0.19%
SAGE
1 publication, 0.19%
World Scientific
1 publication, 0.19%
Bentham Science Publishers Ltd.
1 publication, 0.19%
Cold Spring Harbor Laboratory
1 publication, 0.19%
50
100
150
200
250
300
350
  • We do not take into account publications without a DOI.
  • Statistics recalculated weekly.

Are you a researcher?

Create a profile to get free access to personal recommendations for colleagues and new articles.
Metrics
525
Share
Cite this
GOST |
Cite this
GOST Copy
Peng S. et al. Convolutional Occupancy Networks // Lecture Notes in Computer Science. 2020. pp. 523-540.
GOST all authors (up to 50) Copy
Peng S., Niemeyer M., Mescheder L., POLLEFEYS M., Geiger A. Convolutional Occupancy Networks // Lecture Notes in Computer Science. 2020. pp. 523-540.
RIS |
Cite this
RIS Copy
TY - GENERIC
DO - 10.1007/978-3-030-58580-8_31
UR - https://doi.org/10.1007/978-3-030-58580-8_31
TI - Convolutional Occupancy Networks
T2 - Lecture Notes in Computer Science
AU - Peng, Songyou
AU - Niemeyer, Michael
AU - Mescheder, Lars
AU - POLLEFEYS, MARC
AU - Geiger, Andreas
PY - 2020
DA - 2020/12/02
PB - Springer Nature
SP - 523-540
SN - 0302-9743
SN - 1611-3349
SN - 1861-2075
SN - 1861-2083
ER -
BibTex
Cite this
BibTex (up to 50 authors) Copy
@incollection{2020_Peng,
author = {Songyou Peng and Michael Niemeyer and Lars Mescheder and MARC POLLEFEYS and Andreas Geiger},
title = {Convolutional Occupancy Networks},
publisher = {Springer Nature},
year = {2020},
pages = {523--540},
month = {dec}
}