Video information provides a rich supply of data that's offered to United States nowadays in large quantities e.g. from on-line resources. During this paper present a cosegmentation framework to find and segment out common object regions across multiple frames and multiple videos during a joint fashion. In this paper incorporate 3 kinds of cues, i.e., intraframe salience, interframe consistency, and across-video similarity into an energy optimization framework that doesn't create restrictive assumptions on foreground appearance and motion model, and doesn't need objects to be visible altogether frames. During this paper present a video co-segmentation technique that uses category-independent object proposals as its basic component and may extract multiple foreground objects during a video set.