Object Motion Tracking and Detection in Surveillance Videos using Resnet Architecture

Abstract

Recent years have seen a flurry of activity in the field of security research devoted mostly to the topic of video surveillance. The goal of installing an Intelligent surveillance system is to identify and follow the item in order to foil any malicious intent. Effective methods for using a Video Surveillance System are discussed. Three distinct tasks—Classification, Object Detection, and Tracking—make up the Video Surveillance System's overall workflow. Background Subtraction, Optical Flow, and Spatially Filtering are just few of the approaches that may be used to identify objects in a scene. It is necessary to separate the identified items into distinct categories, which may be done using feature-based methods. Points-, kernel-, and silhouette-based methods may be used to locate the items of interest. Computation time and reliability rates are considered in a comparison of the methods.