NBA SportVU
收藏资源简介:
NBA SportVU数据集是一个用于多对象轨迹预测的扩展数据集,由北京邮电大学网络与交换技术国家重点实验室的研究团队扩展。该数据集包含了篮球比赛中的球员轨迹和团队战术标签,通过人工注释增加了团队级别的战术信息,用于轨迹和战术预测任务。数据集的创建旨在解决复杂体育场景中的轨迹预测问题,特别是篮球比赛中的团队互动和动态意图变化。通过扩展该数据集,研究团队提升了模型在多对象轨迹预测任务中的表现,尤其是在体育分析和决策支持领域。
The NBA SportVU Dataset is an extended dataset for multi-object trajectory prediction, extended by the research team from the State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications. This dataset contains player trajectories and team tactical tags from basketball games, with team-level tactical information added via manual annotation, and is used for trajectory and tactical prediction tasks. The dataset was developed to address trajectory prediction issues in complex sports scenarios, particularly team interactions and dynamic intent changes during basketball matches. By extending this dataset, the research team improved the performance of models in multi-object trajectory prediction tasks, especially in the fields of sports analytics and decision support.




