AVE-Order, ActivityNet-Order
收藏资源简介:
本文介绍了两个视频剪辑中的镜头序列排序任务的新基准数据集:AVE-Order和ActivityNet-Order。AVE-Order基于AVE数据集构建,而ActivityNet-Order基于ActivityNet数据集构建。这两个数据集为公开可用,包含了视频文件和相应的镜头分割,以供研究者准确分析时间关系和叙事逻辑。这些数据集的创建旨在促进视频剪辑中镜头序列排序任务的研究,解决AI辅助视频编辑中的挑战。
This paper presents two novel benchmark datasets for the shot sequence ordering task in video editing: AVE-Order and ActivityNet-Order. AVE-Order is constructed based on the AVE dataset, while ActivityNet-Order is derived from the ActivityNet dataset. Both datasets are publicly available, containing video files and their corresponding shot segmentations to enable researchers to accurately analyze temporal relationships and narrative logic. The creation of these datasets is designed to advance research on the shot sequence ordering task in video editing and resolve the challenges in AI-assisted video editing.




