Generating synthetic LiDAR point cloud data for object detection using the Unreal Game Engine
- Object detection based on artificial intelligence is ubiquitous in today’s computer vision research and application. The training of the neural networks for object detection requires large and high-quality datasets. Besides datasets based on image data, datasets derived from point clouds offer several advantages. However, training datasets are sparse and their generation requires a lot of effort, especially in industrial domains. A solution to this issue offers the generation of synthetic point cloud data. Based on the design science research method, the work at hand proposes an approach and its instantiation for generating synthetic point cloud data based on the Unreal Engine. The point cloud quality is evaluated by comparing the synthetic cloud to a real-world point cloud. Within a practical example the applicability of the Unreal Game engine for synthetic point cloud generation could be successfully demonstrated.
Author: | Mathias EggertORCiD, Maximilian Schade, Florian Bröhl, Alexander Moriz |
---|---|
DOI: | https://doi.org/10.1007/978-3-031-61175-9_20 |
ISBN: | 978-3-031-61174-2 (Print) |
ISBN: | 978-3-031-61175-9 (Online) |
Parent Title (German): | Design Science Research for a Resilient Future (DESRIST 2024) |
Publisher: | Springer |
Place of publication: | Cham |
Editor: | Munir Mandviwalla, Matthias Söllner, Tuure Tuunanen |
Document Type: | Conference Proceeding |
Language: | German |
Year of Completion: | 2024 |
First Page: | 295 |
Last Page: | 309 |
Note: | 19th International Conference on Design Science Research in Information Systems and Technology, DESRIST 2024, Trollhättan, Sweden, June 3–5, 2024 |
Link: | https://doi.org/10.1007/978-3-031-61175-9_20 |
Zugriffsart: | bezahl |
Institutes: | FH Aachen / Fachbereich Wirtschaftswissenschaften |
collections: | Verlag / Springer |