VC NBA 2022
收藏资源简介:
VC NBA 2022是一个专为篮球视频字幕生成设计的多模态数据集,由北京工业大学信息技术学院创建。该数据集包含9种精细化的投篮事件和286名球员的详细信息(如图像和姓名),旨在通过多模态篮球游戏知识图谱(KG NBA 2022)提供视频之外的额外知识。数据集的构建过程涉及从专业篮球平台收集的25场完整NBA比赛数据,包括事件描述、球员和队伍信息及视频。此数据集主要应用于篮球直播文本广播,解决生成包含特定实体名和精细化动作的文本描述问题。
VC NBA 2022 is a multimodal dataset specifically designed for basketball video caption generation, created by the School of Information Technology, Beijing University of Technology. This dataset includes 9 categories of refined shooting events and detailed profiles of 286 players (e.g., their images and names), aiming to provide supplementary knowledge beyond the original video content via the multimodal basketball game knowledge graph KG NBA 2022. The construction of this dataset collected 25 complete NBA game datasets from professional basketball platforms, covering event descriptions, player and team information, as well as video footage. This dataset is primarily utilized for basketball live text broadcasting, to solve the problem of generating text descriptions that contain specific entity names and refined actions.




