PROJECT TITLE :

Video Object Segmentation Via Dense Trajectories

ABSTRACT:

In this paper, we propose a completely unique approach to phase moving object in video by utilizing improved purpose trajectories . First, point trajectories are densely sampled from video and tracked through optical flow, that provides data of long-term temporal interactions among objects in the video sequence . Second, a unique affinity measurement method considering each world and native data of purpose trajectories is proposed to cluster trajectories into teams. Finally, we tend to propose a new graph-based mostly segmentation technique which adopts each native and international motion info encoded by the tracked dense purpose trajectories. The proposed approach achieves good performance on trajectory clustering, and it conjointly obtains correct video object segmentation results on each the Moseg dataset and our new dataset containing additional difficult videos.