A key challenging scenario for autonomous driving is intersections, but there are currently no large-scale public trajectory datasets on signalized intersections. Motivated by this, we constructed the SIND dataset, which was collected at a signalized intersection in Tianjin, China. The SIND dataset is based on 4K video captured by drones, providing information including traffic participant trajectories, traffic light status, and high-definition maps. A demo video of the dataset can be viewed on Youtube or BiliBili. The corresponding paper has been accepted by 2022 IEEE Conference on Intelligent Transportation Systems (ITSC 2022).
SIND contains 7 hours of recording including over 13,000 traffic participants with 7 types, HD maps and traffic light information are used to count traffic light violations by vehicles in them. Clearly, SIND has a high proportion of vulnerable road users and frequent non-motor vehicle violations.
You can get the project and a sample record by executing git clone https://github.com/SOTIF-AVLab/SinD.git. To access the full dataset， please contact us by e-mail:
hong_wang@tsinghua.edu.cn or 13645450063@163.com or 18975505069@163.com
- The title of the email should be: [Apply for SinD] name_country(region)_organization
Please describe your laboratory or department, research interest, and the purpose of the dataset in detail in the email content; After confirmation, we will send the complete data link as soon as possible(By default, a reply email containing a link will be sent to email address which you contact us). You are also welcome to further communicate with us
SIND consists of 23 records, each of which contains 8-22 minutes of trajectories of traffic participants. In addition to the trajectories and motion state parameters of traffic participants, SIND also provides synchronized traffic light states and HD map. Each record contains the following .csv files: For detailed format see Format.md
For visualization see SIND-Vis-tool
The dataset provides a simple traffic light violation label for vehicles. In addition, our team has developed a method to monitor more violations of vehicles
If you use SinD or its code in your work, please cite our datasets as follows:
@INPROCEEDINGS{9921959, author={Xu, Yanchao and Shao, Wenbo and Li, Jun and Yang, Kai and Wang, Weida and Huang, Hua and Lv, Chen and Wang, Hong},
booktitle={2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)},
title={SIND: A Drone Dataset at Signalized Intersection in China},
year={2022}, volume={}, number={}, pages={2471-2478},
doi={10.1109/ITSC55140.2022.9921959}}
Our visualization code is built upon the public code of the following papers:
- The ind dataset: A drone dataset of naturalistic road user trajectories at german intersections, IV'2020
- Constructing a Highly Interactive Vehicle Motion Dataset, IROS'2019
- School of Vehicle and Mobility, Tsinghua University
- Tsinghua Intelligent Vehicle Design and Safety Research Institute
- Safety Of The Intended Functionality（SOTIF） Research Team
- 2022.9.6: Fixed some minor errors. 1.Fixed category: pedestrian 2. Added missing part of traffic light information of 8_5_1
- 2022.9.9: Add paper links and citation requirements in the page
- 2022.10.14: Added item: Application based on this dataset