Search
Classification of objects in the LIDAR point clouds using Deep Neural Networks based on the PointNet model
Abstract
This work attempts to meet the challenges associated with the classification of LIDAR point clouds by means of deep learning. In addition to achieving high accuracy, the designed system should allow the classification of point clouds covering an area of several dozen square kilometers within a reasonable time interval. Therefore, it must be characterized by fast processing and efficient use of memory. Thus, the most popular approaches to the point cloud classification using neural networks are discussed. At the same time, their shortcomings are indicated. A developed model based on the PointNet architecture is presented and the way of preparing data for classification is shown. The model is tested on a cloud coming from the 3D Semantic Labeling competition, achieving a good result, confirmed by the high quality of the system, i.e. a high rate of categorization of objects.
Citations
-
1 3
CrossRef
-
0
Web of Science
-
1 4
Scopus
Authors (2)
Cite as
Full text
- Publication version
- Accepted or Published Version
- License
- Copyright (2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd.)
Keywords
Details
- Category:
- Articles
- Type:
- artykuły w czasopismach
- Published in:
-
IFAC-PapersOnLine
no. 52,
pages 416 - 421,
ISSN: 2405-8963 - Language:
- English
- Publication year:
- 2019
- Bibliographic description:
- Kowalczuk Z., Szymański K.: Classification of objects in the LIDAR point clouds using Deep Neural Networks based on the PointNet model// IFAC-PapersOnLine -Vol. 52,iss. 8 (2019), s.416-421
- DOI:
- Digital Object Identifier (open in new tab) 10.1016/j.ifacol.2019.08.099
- Verified by:
- Gdańsk University of Technology
seen 184 times