本帖最后由 keer_zu 于 2024-3-26 09:43 编辑
Camera-Radar Fusion with BEV Representation for Place Recognition
摘要:相机和雷达数据的互补特征集成已经成为三维目标检测的一种有效方法。
Abstract— The integration of complementary characteristics from camera and radar data has emerged as an effective approach in 3D object detection.
然而,这种基于融合的方法仍未被用于位置识别,而位置识别对于自主系统来说同样重要。
However, such fusion-based methods remain unexplored for place recognition, an equally important task for autonomous systems.
由于位置识别依赖于查询场景和相应候选场景之间的相似性,因此场景的静止背景在任务中起着至关重要的作用。
Given that place recognition relies on the similarity between a query scene and the corresponding candidate scene, the stationary background of a scene is expected to play a crucial role in the task.
因此,目前设计良好的用于三维目标检测的相机-雷达融合方法,由于主要关注动态前景目标,很难有效地进行位置识别。
As such, current well-designed camera-radar fusion methods for 3D object detection can hardly take effect in place recognition because they mainly focus on dynamic foreground objects.
本文提出了一种基于背景关注的摄像机-雷达融合方法CRPlace,从多视角图像和雷达点云中生成背景关注全局描述符,实现精确的位置识别。
In this paper, a background-attentive camera-radar fusion-based method, named CRPlace, is proposed to generate background-attentive global descriptors from multi-view images and radar point clouds for accurate place recognition.
为了有效地提取静止背景特征,我们设计了一个自适应模块,利用相机的BEV特征和雷达的动态点生成背景遮光罩。
To extract stationary background features effectively, we design an adaptive module that generates the backgroundattentive mask by utilizing the camera BEV feature and radar dynamic points.
在背景掩模的引导下,设计了一种基于双向交叉注意力的空间融合策略,实现了相机BEV特征背景信息与雷达BEV特征之间的全面空间交互。
With the guidance of a background mask, we devise a bidirectional cross-attention-based spatial fusion strategy to facilitate comprehensive spatial interaction between the background information of the camera BEV feature and the radar BEV feature.
作为第一个基于摄像头-雷达融合的位置识别网络,CRPlace已经在nuScenes数据集上进行了全面的评估。
As the first camera-radar fusionbased place recognition network, CRPlace has been evaluated thoroughly on the nuScenes dataset.
结果表明,我们的算法在一组综合指标上优于各种基线方法(recall@1达到91.2%)。
The results show that our algorithm outperforms a variety of baseline methods across a comprehensive set of metrics (recall@1 reaches 91.2%).
|