Object detection for large-scale outdoor point cloud based on deep learning

Yaming Wang; Xiaodong Jia; Kui Zhou

doi:10.1117/12.2619490

27 March 2022 Object detection for large-scale outdoor point cloud based on deep learning

Yaming Wang, Xiaodong Jia, Kui Zhou

Proceedings Volume 12169, Eighth Symposium on Novel Photoelectronic Detection Technology and Applications; 1216906 (2022) https://doi.org/10.1117/12.2619490
Event: Eighth Symposium on Novel Photoelectronic Detection Technology and Applications, 2021, Kunming, China

Abstract

A object detection method based on RandLA-Net efficient semantic segmentation network is designed based on the analysis of a variety of current neural networks for real-time obstacle detection during low-altitude flight of helicopters. The method first uses RandLA-Net for semantic segmentation of point clouds, and in order to improve the recognition ability of the network for tall obstacles, the local feature aggregation module in the network is improved accordingly, then the target class of interest is clustered and segmented, and the clustering results are obtained by improving the traditional Euclidean clustering method, and finally the 3D boundingbox of the target is obtained based on the principle of principal component analysis. The paper focuses on the algorithm's ability to identify obstacles such as power lines and tall towers. After actually collecting data and constructing data sets for validation, the final results show that the method can detect targets more accurately, laying a theoretical foundation for application to real-time detection systems.

Citation Download Citation

Yaming Wang, Xiaodong Jia, and Kui Zhou "Object detection for large-scale outdoor point cloud based on deep learning", Proc. SPIE 12169, Eighth Symposium on Novel Photoelectronic Detection Technology and Applications, 1216906 (27 March 2022); https://doi.org/10.1117/12.2619490

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available