Self-Supervised point set local descriptors for Point Cloud Registration
Please always quote using this URN: urn:nbn:de:bvb:20-opus-223000
- Descriptors play an important role in point cloud registration. The current state-of-the-art resorts to the high regression capability of deep learning. However, recent deep learning-based descriptors require different levels of annotation and selection of patches, which make the model hard to migrate to new scenarios. In this work, we learn local registration descriptors for point clouds in a self-supervised manner. In each iteration of the training, the input of the network is merely one unlabeled point cloud. Thus, the whole trainingDescriptors play an important role in point cloud registration. The current state-of-the-art resorts to the high regression capability of deep learning. However, recent deep learning-based descriptors require different levels of annotation and selection of patches, which make the model hard to migrate to new scenarios. In this work, we learn local registration descriptors for point clouds in a self-supervised manner. In each iteration of the training, the input of the network is merely one unlabeled point cloud. Thus, the whole training requires no manual annotation and manual selection of patches. In addition, we propose to involve keypoint sampling into the pipeline, which further improves the performance of our model. Our experiments demonstrate the capability of our self-supervised local descriptor to achieve even better performance than the supervised model, while being easier to train and requiring no data labeling.…
Author: | Yijun Yuan, Dorit BorrmannORCiD, Jiawei Hou, Yuexin Ma, Andreas NüchterORCiD, Sören Schwertfeger |
---|---|
URN: | urn:nbn:de:bvb:20-opus-223000 |
Document Type: | Journal article |
Faculties: | Fakultät für Mathematik und Informatik / Institut für Informatik |
Language: | English |
Parent Title (English): | Sensors |
ISSN: | 1424-8220 |
Year of Completion: | 2021 |
Volume: | 21 |
Issue: | 2 |
Article Number: | 486 |
Source: | Sensors 2021, 21(2), 486; https://doi.org/10.3390/s21020486 |
DOI: | https://doi.org/10.3390/s21020486 |
Dewey Decimal Classification: | 0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 005 Computerprogrammierung, Programme, Daten |
Tag: | descriptors; point cloud registration; self-supervised learning |
Release Date: | 2021/09/30 |
Date of first Publication: | 2021/01/12 |
Open-Access-Publikationsfonds / Förderzeitraum 2021 | |
Licence (German): | CC BY: Creative-Commons-Lizenz: Namensnennung 4.0 International |