CITR dataset 和 DUT dataset

本研究构建了两个数据集：CITR数据集和DUT数据集，用于研究行人与车辆之间的交互作用。CITR数据集通过控制实验设计，专注于基本的车辆-人群交互场景，而DUT数据集则记录了大连理工大学校园内自然发生的车辆-人群交互情况。两个数据集均使用无人机搭载的向下摄像头进行视频录制，提取行人和车辆的轨迹，并通过卡尔曼滤波器进行精炼。这些数据集旨在帮助开发自动驾驶车辆在复杂城市环境中的应对策略，特别是在处理车辆与人群交互时的问题。

In this study, two datasets, namely the CITR Dataset and the DUT Dataset, are constructed to investigate the interactions between pedestrians and vehicles. The CITR Dataset, designed via controlled experiments, focuses on basic vehicle-pedestrian interaction scenarios, while the DUT Dataset records naturally occurring vehicle-pedestrian interactions on the campus of Dalian University of Technology. Both datasets are captured via video recordings using downward-facing cameras mounted on unmanned aerial vehicles (UAVs), with the trajectories of pedestrians and vehicles extracted and refined via Kalman filters. These datasets are intended to facilitate the development of response strategies for autonomous vehicles in complex urban environments, particularly when addressing vehicle-pedestrian interaction challenges.