<- Back to Datasets

D²-City

A Large-scale Video Dataset of Diverse Traffic Scenarios in China

D²-City

The D²-City dataset is a comprehensive collection of dashcam videos collected by vehicles on DiDi’s platform in 5 Chinese cities. All videos are 30-second clips in 720p or 1080p resolution at a frame rate of 25fps. The dataset will be divided into training, validation, and testing subsets. The data files and statistics will then be released in stages.For around 1000 of the videos, we have annotated the bounding boxes and tracking IDs of road objects into 12 different categories, shown in the following table. For some of the remainder of the videos, we annotate the bounding boxes in key frames.

View this Dataset
->
View author website
Task
Autonomous Driving
Annotation Types
Bounding Boxes
10000
Items
12
Classes
10000
Labels
Models using this dataset
Last updated on 
January 20, 2022
Licensed under 
Research Only
Gain control of your training data
15,000+ ML engineers can’t be wrong