ROCO
收藏资源简介:
ROCO数据集是由中国科学技术大学和美国密歇根大学的研究团队共同创建,专注于环形交叉口的交通冲突研究。该数据集包含557个交通冲突样本和17个交通碰撞样本,数据来源于安装在环形交叉口的四个鱼眼摄像头捕捉的视频流。数据集的创建过程涉及使用深度学习算法自动识别潜在的交通冲突,随后进行人工标注和轨迹数据提取。ROCO数据集的应用领域主要集中在交通安全的改进,特别是在环形交叉口的交通冲突分析和预测,旨在通过数据分析指导交通管理和设计,减少交通事故的发生。
The ROCO dataset was co-created by research teams from the University of Science and Technology of China and the University of Michigan, focusing on traffic conflict research at roundabouts. This dataset contains 557 traffic conflict samples and 17 traffic collision samples, with data derived from video streams captured by four fisheye cameras deployed at roundabouts. The dataset creation process involves automatically identifying potential traffic conflicts using deep learning algorithms, followed by manual annotation and trajectory data extraction. The primary application areas of the ROCO dataset center on traffic safety improvement, specifically traffic conflict analysis and prediction at roundabouts, aiming to guide traffic management and design via data analysis to reduce the incidence of traffic accidents.




