VISIOCITY

VISIOCITY是一个大规模、多样化的视频摘要数据集，由印度理工学院孟买分校创建。该数据集包含67个视频，涵盖六个不同类别：电视节目、体育、监控、教育、生日和婚礼。视频平均时长约50分钟，具有密集的概念标注，支持不同类型的视频摘要和计算机视觉任务。数据集的创建旨在解决现有数据集视频长度短、类型单一的问题，为自动视频摘要技术提供更真实、多样的测试平台。VISIOCITY的应用领域广泛，旨在通过提供高质量、非冗余的短视频摘要，帮助用户快速获取视频中的关键信息，适用于娱乐、教育、监控等多个领域。

VISIOCITY is a large-scale, diverse video summarization dataset developed by the Indian Institute of Technology Bombay. This dataset comprises 67 videos spanning six distinct categories: television programs, sports, surveillance, education, birthday, and weddings. The average duration of the videos is approximately 50 minutes, and it features dense conceptual annotations that support various video summarization and computer vision tasks. The dataset was created to address the shortcomings of existing datasets, which usually have short video lengths and limited category diversity, providing a more realistic and diverse testbed for automatic video summarization technologies. VISIOCITY has broad application prospects: it aims to help users quickly acquire key information from videos by providing high-quality, non-redundant short video summaries, making it applicable to multiple fields including entertainment, education, and surveillance.