Point Cloud Semantic Segmentation Method Based on Multi-scale Supervision

DOI

10.2991/978-94-6239-774-3_33How to use a DOI?

Keywords

Point cloud semantic segmentation; Multi-scale supervision; Contrastive learning

Abstract

Due to the ability to faithfully reconstruct real-world 3D scenes, point cloud data are widely used in fields such as digital twins and scene understanding. However, manual annotation of large-scale point clouds is inefficient, whereas point-cloud semantic segmentation technology can extract the geometric features of various objects in 3D scenes to generate fine-grained masks for different categories. To simulate a real air combat environment, we constructed safe test conditions on the ground and used LIDAR to scan the fighter jet cockpit. For the precise differentiation of the various avionics systems in the virtual cockpit, we designed a point cloud semantic segmentation network based on multi-scale supervision, which employs an encoder to generate multi-hot labels for supervising feature maps at corresponding scales across multiple decoders. We performed 6-fold cross-validation of our network on the indoor point cloud dataset S3DIS and transferred the model to cockpit point cloud data for segmentation. Ultimately, we achieved an mIoU of 75.3% and an OA of 90.2% on S3DIS.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Zhiyuan Wang
AU  - Shaoqing Liu
AU  - Wei Jiang
AU  - Jiaxu Wang
PY  - 2026
DA  - 2026/09/11
TI  - Point Cloud Semantic Segmentation Method Based on Multi-scale Supervision
BT  - Proceedings of the 2026 5th International Conference on Mathematical Statistics and Economic Analysis (MSEA 2026
PB  - Atlantis Press
SP  - 348
EP  - 357
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6239-774-3_33
DO  - 10.2991/978-94-6239-774-3_33
ID  - Wang2026
ER  -