Bi-Att3DDet: Attention-Based Bi-Directional Fusion for Multi-Modal 3D Object Detection
Sensors · 23.01.2025 · 25 (3) · 658
DOI: 10.3390/s25030658 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 23.01.2025. Verfasst von Xu Gao und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer and Artificial Intelligence, Zhengzhou University, Zhengzhou 450001, China. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Zhengzhou Major Science and Technology Project, Technology Special Projects in Henan Province.
Abstract (Originalsprache)
Currently, multi-modal 3D object detection methods have become a key area of research in the field of autonomous driving. Fusion is an essential factor affecting performance in multi-modal object detection. However, previous methods still suffer from the inability to effectively fuse features from LiDAR and RGB images, resulting in a low utilization rate of complementary information between depth and semantic texture features. At the same time, existing methods may not adequately capture the structural information in Region of Interest (RoI) features when extracting them. Structural information plays a crucial role in RoI features. It encompasses the position, size, and orientation of objects, as well as the relative positions and spatial relationships between objects. Its absence can result in false or missed detections. To solve the above problems, we propose a multi-modal sensor fusion network, Bi-Att3DDet, which mainly consists of a Self-Attentive RoI Feature Extraction module (SARoIFE) and a Feature Bidirectional Interactive Fusion module (FBIF). Specifically, SARoIFE captures the relationship between different positions in RoI features to obtain high-quality RoI features through the self-attention mechanism. SARoIFE prepares for the fusion stage. FBIF performs bidirectional interaction between LiDAR and pseudo RoI features to make full use of the complementary information. We perform comprehensive experiments on the KITTI dataset, and our method notably demonstrates a 1.55% improvement in the hard difficulty level and a 0.19% improvement in the mean Average Precision (mAP) metric on the test dataset.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Computer and Artificial Intelligence, Zhengzhou University, Zhengzhou 450001, China.
- National Supercomputing Center in Zhengzhou, Zhengzhou 450001, China.
- Yutong Bus Co., Ltd., Zhengzhou 450000, China.
ORCID-Profile: Xu Gao, Yaqian Zhao, Yanan Wang, Gang Wu
Förderung
- Zhengzhou Major Science and Technology Project
- Technology Special Projects in Henan Province
Zitieren
APA
Gao, X., Zhao, Y., Wang, Y., Shang, J., Zhang, C., & Wu, G. (2025). Bi-Att3DDet: Attention-Based Bi-Directional Fusion for Multi-Modal 3D Object Detection. Sensors, 25(3), 658. https://doi.org/10.3390/s25030658
Vancouver
Gao X, Zhao Y, Wang Y, Shang J, Zhang C, Wu G. Bi-Att3DDet: Attention-Based Bi-Directional Fusion for Multi-Modal 3D Object Detection. Sensors. 2025;25(3):658. doi:10.3390/s25030658
BibTeX
@article{Gao2025,
author = {Gao, Xu and Zhao, Yaqian and Wang, Yanan and Shang, Jiandong and Zhang, Chunmin and Wu, Gang},
title = {Bi-Att3DDet: Attention-Based Bi-Directional Fusion for Multi-Modal 3D Object Detection},
journal = {Sensors},
year = {2025},
volume = {25},
number = {3},
pages = {658},
doi = {10.3390/s25030658}
}
EUVIMED