RadarScenes Radar Point Cloud Dataset

RadarScenes is a real-world radar point cloud dataset for automotive applications, consisting of 4 hours of driving measurements and point-by-point annotations. The data is collected by 4 series of radar sensors on the test car, and more than 7,500 individual detections of dynamic objects are manually grouped and annotated. The purpose of this dataset is to develop new machine learning-based radar perception algorithms with a focus on mobile road users.