EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Mamba-BEV: A Multiscale State-Space Framework for 3D Object Detection from Point Clouds

Yuyang Liu, Jiabin Wang, Min Mao, Kun Zhang, Yu Xu iD, Mingchen Zhu, Xianjun Wu

Sensors · 20.08.2026 · 26 (16) · 5271

DOI: 10.3390/s26165271 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 20.08.2026. Verfasst von Yuyang Liu und 6 weiteren Autor:innen. Beteiligte Einrichtung: College of Physics and Electronic Engineering, Xinyang Normal University, Xinyang 464000, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Xinyang Normal University.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

LiDAR point clouds are sparse, irregular, and unevenly distributed, which makes representative feature extraction challenging for 3D object detection. Currently, Mamba modules have been increasingly applied to 3D object detection due to their ability to efficiently model global spatial dependencies. However, preserving geometric structures while modeling global spatial dependencies remains challenging in Mamba-based frameworks. In view of this, this paper proposes a single-stage 3D object detection framework for LiDAR-based point clouds. First, this paper designs a hierarchical multiscale structure called Multiscale Voxel–Point Alternating Fusion (MVPF) Module. Within this module, the 2DMamba module is introduced into the feature extraction stage at each scale to model global spatial dependencies in the BEV plane through selective state-space scanning. Second, we design a Z-to-Channel cross-dimensional reorganization strategy that merges the voxel Z dimension into the channel dimension, yielding a BEV-form feature representation suitable for 2DMamba processing. Third, this paper proposes a Local Voxel Feature Enhancement (LVFE) module composed of a Point-to-Voxel Feature Aggregation (PVFA) module, a Voxel Densification Module (VDM), and a coordinate-indexed Voxel-to-Point Mapping (VPM) module, which enhances local voxel representations while maintaining point–voxel spatial correspondence. On the KITTI validation set, Mamba-BEV achieved higher detection accuracy and inference speed than the baseline. Under the Moderate difficulty level of APR11, the 3D AP for Car, Pedestrian, and Cyclist improved by 0.6%, 1.5%, and 0.4%, respectively, compared to the baseline, while the inference speed increased from 44.05 FPS to 67.11 FPS. These results demonstrate the effectiveness of the proposed Mamba-BEV for point cloud-based 3D object detection.

Schlagwörter

Stichwörter der Autor:innen

3D object detectionMambabird’s-eye view (BEV)point cloudstate-space model

Autor:innen und Institutionen

  • College of Physics and Electronic Engineering, Xinyang Normal University, Xinyang 464000, China.
  • School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China.
  • Xinyang Gumai Optronics Co., Ltd., Xinyang 464000, China.
  • School of Geomatics Science and Technology, Nanjing Tech University, Nanjing 210009, China.
  • Xinyang Central Semiconductor Technology Co., Ltd., Xinyang 464000, China.

ORCID-Profile: Yu Xu

Förderung

  • Xinyang Normal University

Zitieren

APA

Liu, Y., Wang, J., Mao, M., Zhang, K., Xu, Y., Zhu, M., & Wu, X. (2026). Mamba-BEV: A Multiscale State-Space Framework for 3D Object Detection from Point Clouds. Sensors, 26(16), 5271. https://doi.org/10.3390/s26165271

Vancouver

Liu Y, Wang J, Mao M, Zhang K, Xu Y, Zhu M, et al. Mamba-BEV: A Multiscale State-Space Framework for 3D Object Detection from Point Clouds. Sensors. 2026;26(16):5271. doi:10.3390/s26165271

BibTeX
@article{Liu2026,
  author = {Liu, Yuyang and Wang, Jiabin and Mao, Min and Zhang, Kun and Xu, Yu and Zhu, Mingchen and Wu, Xianjun},
  title = {Mamba-BEV: A Multiscale State-Space Framework for 3D Object Detection from Point Clouds},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {16},
  pages = {5271},
  doi = {10.3390/s26165271}
}
Kostenlos starten →