Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | 2020 IEEE International Conference on Robotics and Automation, ICRA 2020 |
| Herausgeber (Verlag) | Institute of Electrical and Electronics Engineers Inc. |
| Seiten | 3161-3167 |
| Seitenumfang | 7 |
| ISBN (elektronisch) | 9781728173955 |
| ISBN (Print) | 978-1-7281-7394-8, 978-1-7281-7396-2 |
| Publikationsstatus | Veröffentlicht - 2020 |
| Veranstaltung | 2020 IEEE International Conference on Robotics and Automation, ICRA 2020 - Paris, Frankreich Dauer: 31 Mai 2020 → 31 Aug. 2020 |
Publikationsreihe
| Name | Proceedings - IEEE International Conference on Robotics and Automation |
|---|---|
| ISSN (Print) | 1050-4729 |
Abstract
The horizon line is an important geometric feature for many image processing and scene understanding tasks in computer vision. For instance, in navigation of autonomous vehicles or driver assistance, it can be used to improve 3D reconstruction as well as for semantic interpretation of dynamic environments. While both algorithms and datasets exist for single images, the problem of horizon line estimation from video sequences has not gained attention. In this paper, we show how convolutional neural networks are able to utilise the temporal consistency imposed by video sequences in order to increase the accuracy and reduce the variance of horizon line estimates. A novel CNN architecture with an improved residual convolutional LSTM is presented for temporally consistent horizon line estimation. We propose an adaptive loss function that ensures stable training as well as accurate results. Furthermore, we introduce an extension of the KITTI dataset which contains precise horizon line labels for 43699 images across 72 video sequences. A comprehensive evaluation shows that the proposed approach consistently achieves superior performance compared with existing methods.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Software
- Ingenieurwesen (insg.)
- Steuerungs- und Systemtechnik
- Informatik (insg.)
- Artificial intelligence
- Ingenieurwesen (insg.)
- Elektrotechnik und Elektronik
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
2020 IEEE International Conference on Robotics and Automation, ICRA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. S. 3161-3167 9197170 (Proceedings - IEEE International Conference on Robotics and Automation).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Temporally Consistent Horizon Lines
AU - Kluger, Florian
AU - Ackermann, Hanno
AU - Ying Yang, Michael
AU - Rosenhahn, Bodo
N1 - Funding information: Acknowledgement: This work was supported by German Research Foundation (DFG) grant Ro 2497 / 12-2.
PY - 2020
Y1 - 2020
N2 - The horizon line is an important geometric feature for many image processing and scene understanding tasks in computer vision. For instance, in navigation of autonomous vehicles or driver assistance, it can be used to improve 3D reconstruction as well as for semantic interpretation of dynamic environments. While both algorithms and datasets exist for single images, the problem of horizon line estimation from video sequences has not gained attention. In this paper, we show how convolutional neural networks are able to utilise the temporal consistency imposed by video sequences in order to increase the accuracy and reduce the variance of horizon line estimates. A novel CNN architecture with an improved residual convolutional LSTM is presented for temporally consistent horizon line estimation. We propose an adaptive loss function that ensures stable training as well as accurate results. Furthermore, we introduce an extension of the KITTI dataset which contains precise horizon line labels for 43699 images across 72 video sequences. A comprehensive evaluation shows that the proposed approach consistently achieves superior performance compared with existing methods.
AB - The horizon line is an important geometric feature for many image processing and scene understanding tasks in computer vision. For instance, in navigation of autonomous vehicles or driver assistance, it can be used to improve 3D reconstruction as well as for semantic interpretation of dynamic environments. While both algorithms and datasets exist for single images, the problem of horizon line estimation from video sequences has not gained attention. In this paper, we show how convolutional neural networks are able to utilise the temporal consistency imposed by video sequences in order to increase the accuracy and reduce the variance of horizon line estimates. A novel CNN architecture with an improved residual convolutional LSTM is presented for temporally consistent horizon line estimation. We propose an adaptive loss function that ensures stable training as well as accurate results. Furthermore, we introduce an extension of the KITTI dataset which contains precise horizon line labels for 43699 images across 72 video sequences. A comprehensive evaluation shows that the proposed approach consistently achieves superior performance compared with existing methods.
UR - http://www.scopus.com/inward/record.url?scp=85092732502&partnerID=8YFLogxK
U2 - 10.1109/ICRA40945.2020.9197170
DO - 10.1109/ICRA40945.2020.9197170
M3 - Conference contribution
AN - SCOPUS:85092732502
SN - 978-1-7281-7394-8
SN - 978-1-7281-7396-2
T3 - Proceedings - IEEE International Conference on Robotics and Automation
SP - 3161
EP - 3167
BT - 2020 IEEE International Conference on Robotics and Automation, ICRA 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2020 IEEE International Conference on Robotics and Automation, ICRA 2020
Y2 - 31 May 2020 through 31 August 2020
ER -