RadarScenes

RadarScenes 是用于汽车应用的真实世界雷达点云数据集。它由安装在一辆测试车辆上的四个系列雷达传感器收集的四个多小时驾驶的测量值和逐点注释组成。动态对象的单个检测被手动分组到集群并随后标记。该数据集的目的是开发新的（基于机器学习的）雷达感知算法，重点是移动道路使用者。记录序列的图像是使用纪实相机拍摄的。

RadarScenes is a real-world radar point cloud dataset tailored for automotive applications. It consists of measurements and per-point annotations collected over four hours of driving by four series radar sensors installed on a test vehicle. Individual detections of dynamic objects are manually grouped into clusters and subsequently labeled. The core objective of this dataset is to facilitate the development of novel machine learning-based radar perception algorithms, with a primary focus on moving road users. Images of the driving recording sequences were captured using a documentary camera.

搜集汇总

背景与挑战

背景概述

RadarScenes是一个专为汽车应用设计的真实世界雷达点云数据集，包含超过四小时驾驶数据，由四个雷达传感器采集，并提供了逐点注释和动态对象的手动分组标签，旨在支持移动道路使用者的雷达感知算法开发。该数据集由多特蒙德工业大学和梅赛德斯奔驰于2021年发布，采用CC BY-NC-SA 4.0许可证，数据规模为10.7GB，适用于点云目标检测任务。