BOVText

BOVText是一个大规模的双语视频文本数据集，由浙江大学创建，包含2021个视频，覆盖30多个开放类别，如生活Vlog、驾驶、电影等。数据集提供了超过1,750,000帧的视频内容，支持丰富的文本类型标注，如标题、字幕或场景文本，并提供中英文双语文本标注，以促进多文化交流。BOVText旨在解决视频文本识别和跟踪的挑战，适用于视频理解、视频检索和视频文本翻译等领域。

BOVText is a large-scale bilingual video-text dataset created by Zhejiang University. It comprises 2021 videos covering more than 30 open categories such as lifestyle Vlogs, driving, movies and other types. The dataset provides over 1,750,000 video frames, supports rich annotations of diverse text types including titles, subtitles and scene text, and offers Chinese-English bilingual text annotations to facilitate cross-cultural communication. BOVText aims to address the challenges in video text recognition and tracking, and is applicable to fields such as video understanding, video retrieval and video-text translation.