A Graph-Based Approach for Estimating Object Interactions in Surveillance Video Synopsis

Video synopsis is a technique that condenses surveillance videos into compact equivalent videos. The objective of this paper is to identify and group related moving objects (also referred to as tubes) in a video using a graph-based video synopsis method. To do this, we first develop an interaction matrix to determine the degree of pairwise interaction between the tubes. A tube-set generation graph is then suggested as a way to combine linked tubes into a single tube set. Publicly available video datasets have been utilized by us to demonstrate the effectiveness of our method. Our technique has come up with another way of video synopsis; that is, focusing on the context of the original movie and picking out important details. This makes it a reliable surveillance approach as it greatly improves security efficiency through fast and accurate compression of videos. Moreover, our method grounded in graph preservation eases our understanding of what connects various elements in a film or video.