This article is an original article of the blogger, and may not be reproduced without the permission of the blogger.
This article is a series of articles in the column "python 3D point cloud from basics to deep learning", and the address is "https://blog.csdn.net/suiyingy/article/details/124017716".
NuScenes has a lot of introduction materials on the Internet, but most of them are just translations of the official website, lacking the internal connection introduction of each file. For example, the data format of nuScenes lidar, which attributes the point cloud contains. Another example is the file relationship between the sample folder and the sweeps folder, and how it is reflected in the relevant json file. This article will detail the various parts of the nuScenes dataset.
1 basic information of nuScenes
The nuScenes dataset is a public large-scale dataset for autonomous driving developed by the Motional (formerly nuTonomy) team. The data set comes from 1,000 driving scenes collected in Boston and Singapore. Each scene selects a 20-second long video, with a total of about 15 hours of driving data. When selecting scenes, fully consider diverse driving operations, traffic conditions and unexpected situations, such as different locations, weather conditions, vehicle types, vegetation, road signs and driving rules, etc. Compared with the Kitti dataset, nuScenes is more complex and richer in scenes.
The nuScenes dataset was officially released in March 2019. The full dataset includes approximately 1.4 million images, 390,000 lidar point clouds, 1.4 million radar scans, and 1.4 million object bounding boxes in 40,000 keyframes. nuScenes is