Detecting intended human objects in human-captured videos

When people take videos, they always want to capture intended objects, which are essential for presenting what they want to express in their videos, and to share the intended objects with others. The concept of intended objects provide a novel perspective for video content analysis, and detecting intended objects may be beneficial for wide range of applications such as video understanding and semantics interpretation, video summarization, video adaptation, video privacy protection, and so on. In this paper, we focus on a particular type of intended objects, i.e., intended human objects, and an interesting method is developed for detecting intended human objects automatically from human-captured videos. We also investigate the correlation between intended human objects and visual attention. Our experimental results indicate that our method can successfully detect the intended human objects.