VSCrowd
收藏资源简介:
VSCrowd是一个大规模的视频人群定位数据集,由墨尔本大学计算与信息系统学院的研究团队创建。该数据集包含超过60,000帧高清监控视频,覆盖多种场景,如商场、街道和景点,共有超过234万个头部标注。数据集不仅提供了头部中心的点标注,还包括每个个体的边界框和轨迹信息,适用于视频人群定位和计数等任务。VSCrowd数据集的创建旨在推动视频人群分析技术的发展,特别是在复杂场景下的人群定位和行为理解。
VSCrowd is a large-scale video crowd localization dataset created by the research team from the School of Computing and Information Systems at the University of Melbourne. This dataset contains over 60,000 frames of high-definition surveillance videos, covering various scenarios including shopping malls, streets and tourist attractions, with more than 2.34 million head annotations. Besides the point annotations of head centers, it also provides bounding boxes and trajectory information for each individual, which is suitable for tasks such as video crowd localization and counting. The creation of the VSCrowd dataset aims to advance the development of video crowd analysis technologies, particularly crowd localization and behavior understanding in complex scenarios.



