D2-City

D2-City是由滴滴出行人工智能实验室创建的大规模行车记录视频数据集，包含超过11,211个视频片段，总时长约一百小时，涵盖中国多个城市的真实交通场景。数据集通过滴滴平台上的车辆收集，特别关注不同天气、道路和交通条件下的数据多样性。创建过程中，数据经过严格筛选，确保质量和场景多样性，同时保护隐私。D2-City旨在推动智能驾驶领域的感知和场景理解研究，解决复杂交通环境下的检测和跟踪问题。

D2-City is a large-scale driving recording video dataset developed by the DiDi Chuxing Artificial Intelligence Lab. It comprises over 11,211 video clips with a total duration of around 100 hours, encompassing real-world traffic scenarios across multiple cities in China. The dataset is collected from vehicles operating on the DiDi platform, with particular focus on data diversity across diverse weather, road and traffic conditions. During the dataset's creation, rigorous filtering was applied to ensure data quality and scene diversity, while also protecting user privacy. D2-City is intended to promote research on perception and scene understanding in the intelligent driving domain, and resolve detection and tracking challenges in complex traffic environments.