论文阅读 【CVPR-2022】 A VERSATILE MULTI-VIEW FRAMEWORK FOR LIDAR-BASED 3D OBJECT DETECTION WITH GUIDANC

论文阅读 【CVPR-2022】 A VERSATILE MULTI-VIEW FRAMEWORK FOR LIDAR-BASED 3D OBJECT DETECTION WITH GUIDANCE FROM PANOPTIC SEGMENTATION

基于激光雷达的以全景分割为指导的三维物体检测的多功能多视角框架

studyai.com

搜索论文: A VERSATILE MULTI-VIEW FRAMEWORK FOR LIDAR-BASED 3D OBJECT DETECTION WITH GUIDANCE FROM PANOPTIC SEGMENTATION

摘要(Abstract)

3D object detection using LiDAR data is an indispensable component for autonomous driving systems.
Yet, only a few LiDAR-based 3D object detection methods leverage segmentation information to further guide the detection process.

使用LiDAR数据的3D物体检测是自主驾驶系统不可或缺的组成部分。然而，只有少数基于LiDAR的三维物体检测方法利用分割信息来进一步引导检测过程。

In this paper, we propose a novel multi-task framework that jointly performs 3D object detection and panoptic segmentation.

在本文中，我们提出了一个新颖的多任务框架，可以联合进行三维物体检测和全景分割。

In our method, the 3D object detection backbone in Bird’s-Eye-View (BEV) plane is augmented by the injection of Range-View (RV) feature maps from the 3D panoptic segmentation backbone.

在我们的方法中，鸟瞰图（BEV）平面上的三维物体检测主干网络被来自三维全景分割主干网络的Range-View（RV）特征图的注入所增强。

This enables the detection backbone to leverage multi-view information to address the shortcomings of each projection view.

这使检测主干能够利用多视图信息来解决每个投影视图的缺点。

Furthermore, foreground semantic information is incorporated to ease the detection task by highlighting the locations of each object class in the feature maps.

此外，通过突出每个物体类别在特征图中的位置把前景语义信息纳入其中，使得检测任务更容易进行。

Finally, a new center density heatmap generated based on the instance-level information further guides the detection backbone by suggesting possible box center locations for objects.

最后，根据实例级信息生成的新的中心密度热图，通过提示物体可能的箱体中心位置，进一步指导检测骨干网络。

Our method works with any BEV-based 3D object detection method, and as shown by extensive experiments on the nuScenes dataset, it provides significant performance gains.

我们的方法适用于任何基于鸟瞰图(BEV)的三维物体检测方法，正如在nuScenes数据集上进行的大量实验所显示的，它提供了显著的性能提升。

Notably, the proposed method based on a single-stage CenterPoint 3D object detection network achieved state-of-the-art performance on nuScenes 3D Detection Benchmark with 67.3 NDS.

值得注意的是，在nuScenes 3D检测基准上，本文提出的基于单阶段CenterPoint 3D物体检测网络的方法取得了最先进的性能，NDS为67.3。