ZJUODset: a long range 3D object detection dataset with 4D radar

In the field of autonomous driving, 3D object detection has consistently been an active topic. Most of existing approaches for the detection utilize images, LiDAR point clouds or fusion of them. The emergence of 4D millimeter wave radar presents a novel solution for this task. It can provide 3D point clouds as well as doppler velocity for the surrounding environment. The long range of radar can benefit the detection of targets in longer distance. However, there are limited existing datasets involving 4D radar. The accurate annotation of distant 3D objects is also a problem. In this paper, we present a novel dataset containing dense LiDAR, image and 4D millimeter wave radar named ZJUODset and a baseline method for 3D object detection based on our dataset. The dataset will be available at https://github.com/Ruoyu-Xu/ZJUODset.