YouTube Traffic Accident Videos

该数据集包含了从YouTube收集的29段短视频，内容涵盖了各种轨迹冲突案例，包括车辆与车辆（V2V）、车辆与自行车（V2B）以及车辆与行人（V2P）的交互场景。这些视频涵盖了不同的天气和光照条件。每段视频剪辑都包含了轨迹冲突前后几秒钟的内容，具有1280×720像素的空间分辨率和每秒30帧的帧率。该数据集的规模为29段短视频，其任务是对交通事故进行检测以及对轨迹冲突进行分析。

This dataset consists of 29 short videos collected from YouTube, encompassing a wide range of trajectory conflict scenarios including vehicle-to-vehicle (V2V), vehicle-to-bicycle (V2B), and vehicle-to-pedestrian (V2P) interaction cases. The videos cover diverse weather and lighting conditions. Each video clip retains several seconds of footage both before and after the occurrence of a trajectory conflict, with a spatial resolution of 1280×720 pixels and a frame rate of 30 frames per second (fps). The primary tasks supported by this dataset are traffic accident detection and trajectory conflict analysis.