Bi-Level Routing and Sparse Spatial Attention based Multi-View BEV 3D Object Detection for Autonomous Driving

Jing Zhang; Jiaqi Liu; Zibo Wang

cs.CVSep 12, 2026

Bi-Level Routing and Sparse Spatial Attention based Multi-View BEV 3D Object Detection for Autonomous Driving

Authors: Jing Zhang, Jiaqi Liu, Zibo Wang

Organizations: Stuart Weitzman School of Design University of Pennsylvania Philadelphia, USA · Courant Institute of Mathematical Science New York University New York, USA

Abstract

Bird's Eye View (BEV)-based multi-view 3D object detection suffers from challenges of computational complexity, multi-scale feature extraction, and efficiency of dense 2D-to-BEV view transformation. To address these problems, this paper proposes an improved BEV 3D object detection algorithm Sparse-BEVNet. Firstly, a Bi-Level Routing Attention (BRA) mechanism is introduced into the image feature extraction network to reduce the computational burden of the backbone. Second, Cascaded Group Attention (CGA) is employed in the feature fusion module, which enhances deep interaction across features of different hierarchical levels without introducing additional computational overhead. Furthermore, a Sparse Spatial Cross-Attention mechanism is adopted to replace the conventional dense view projection pipeline. Experimental results on the public nuScenes dataset demonstrate that the proposed method achieves a mean Average Precision (mAP) of 45.2% and a nuScenes Detection Score (NDS) of 54.5%, corresponding to 3.6% and 2.8% improvements relative to the baseline model, respectively.

Figures & tables

Fig. 1: Overall architecture of the proposed Sparse-BEVNet.

Method	Backbone	mAP (%)	NDS (%)	FPS
FCOS3D	ResNet-101	29.5	37.2	12
DETR3D	ResNet-101	30.3	37.4	14
BEVFormer-S	ResNet-101	41.6	51.7	15
Ours	BiFormer	45.2	54.5	22

TABLE I: Performance comparison of different multi-view 3D object detection algorithms on the nuScenes validation set

BRA	CGA	SSCA	mAP (%)	NDS (%)	FPS
			41.6	51.7	15
$\checkmark$			42.8	52.6	17
$\checkmark$	$\checkmark$		43.7	53.4	16
$\checkmark$	$\checkmark$	$\checkmark$	45.2	54.5	22

TABLE II: Ablation results of core components in Sparse-BEVNet

Fusion Resolution Tiers	mAP (%)	NDS (%)
Deep layers only $\{C_{4},C_{5}\}$	40.8	51.0
Partial cascade $\{C_{3},C_{4},C_{5}\}$	43.1	52.8
Full cascade $\{C_{2},C_{3},C_{4},C_{5}\}$ (Proposed)	45.2	54.5

TABLE III: Comparison of alternative multi-scale merging paths in the CGA neck

Multi-View 3D Object Detection

Multi-View 3D Object Detection combines images from multiple cameras to locate and classify objects in three-dimensional space.

See all 25 papers

Radar-Camera BEV Multi-Task Learning with Cross-Task Attention Bridge for Joint 3D Detection and Segmentation
Apr 14, 2026Ahmet İnanç, Özgür ErkentAutonomous Driving Perception Multi-Task Learning
Hacettepe University, Ankara, Turkey.
Sparse Recovery Bird's-Eye View Model Soup for Robust Camera-Based 3D Object Detection
Oct 26, 2026Jinwoo Chung, Sangho An, Hyunsung Park +1Corruption Robustness 3D Object Detection
Kookmin University, Seoul, Republic of Korea · Hyundai Motor Company · Corresponding author: Jangho Kim
Fusion4CA: Boosting 3D Object Detection via Comprehensive Image Exploitation
Mar 5, 2026Kang Luo, Xin Chen, Yangyi Xiao +13D Object Detection Camera-LiDAR Fusion
IRMV Lab, the Department of Automation, Shanghai Jiao Tong University.

Apr 14, 2026J/K move · Enter open · S save

Radar-Camera BEV Multi-Task Learning with Cross-Task Attention Bridge for Joint 3D Detection and Segmentation

Ahmet İnanç, Özgür Erkent

Hacettepe University, Ankara, Turkey.

Autonomous Driving Perception Multi-Task Learning 3D Object Detection 3D Semantic Segmentation Camera-Radar Fusion Cross-Attention

Open paper AskPDF arXiv

Bird's-eye-view (BEV) representations are the dominant paradigm for 3D perception in autonomous driving, providing a unified spatial canvas where detection and segmentation features are geometrically registered to the same physical coordinate system. However, existing radar-camera fusion methods treat these tasks in isolation, missing the opportunity for cross-task feature sharing: object-level geometric cues from detection can sharpen segmentation, while dense road-layout context from segmentation can anchor detection. We propose CTAB (Cross-Task Attention Bridge), a bidirectional module that exchanges features between detection and segmentation branches via multi-scale deformable attention in shared BEV space. CTAB is integrated into a multi-task framework with an Instance Normalization-based segmentation decoder and learnable BEV upsampling to provide a more detailed BEV representation. On nuScenes, CTAB improves segmentation on 7 classes over the joint multi-task baseline at essentially neutral detection. On a 4-class subset (drivable area, pedestrian crossing, walkway, vehicle), our joint multi-task model achieves 51.0 mIoU-4 while simultaneously providing competitive 3D detection.

Abstract

Multi-View 3D Object Detection

Explore similar work

Radar-Camera BEV Multi-Task Learning with Cross-Task Attention Bridge for Joint 3D Detection and Segmentation

Sparse Recovery Bird's-Eye View Model Soup for Robust Camera-Based 3D Object Detection

Fusion4CA: Boosting 3D Object Detection via Comprehensive Image Exploitation