VOS
收藏资源简介:
VOS是一个包含200个视频的大型视频显著性检测数据集,总时长64分钟。该数据集通过手动标注7650个关键帧中的所有对象和区域,并收集23名受试者的眼动追踪数据来构建。VOS旨在通过结合手动标注的对象/区域掩码和多受试者的眼动追踪数据,明确地定义和标注视频中的显著对象。数据集的应用领域是视频显著性对象检测,旨在解决视频中显著对象的自动检测和分割问题。
VOS is a large-scale video saliency detection dataset consisting of 200 videos with a total duration of 64 minutes. This dataset is constructed by manually annotating all objects and regions across 7650 key frames, and collecting eye-tracking data from 23 participants. VOS aims to explicitly define and annotate salient objects in videos by combining manually annotated object/region masks and eye-tracking data from multiple participants. The dataset targets the domain of video salient object detection, with the goal of addressing the automatic detection and segmentation of salient objects in videos.




