There is an exponential rise in the demand for automatic methods for analyzing the enormous quantities of surveillance video data generated perennially by closed-circuit television (CCTV) systems. One of the main objectives of deploying an automated visual surveillance system is to detect anomalous behavior patterns and recognize the normal ones. Lot of work has been done in this field using object tracking techniques which deploy kalman or particle filter to serve the purpose. The particle filters show good results in Nonlinear Non-Gaussian environment, thus overcoming the limitation of kalman filters but they fail to track occluded target. Thus, basic particle filters give poor results in tracking objects in crowded video sequences where there is a possibility of the target object to get occluded. The current study describes a new method of tracking an object in a video using Branching Particle filter for providing precision in tracking thereby increasing the accuracy in detecting anomalies in crowded videos.