Date of Publication :28th April 2023

Abstract: One of the most crucial jobs in computer vision is object tracking, which has many real-world uses in fields like robotics, automated car tracking, and traffic monitoring. Using the camera in video sequences, object tracking is the proceeding of finding moving things over time. Object recognition is the first step in tracking and aims to identify or find the moving object in the frame. Following that, the observed object can be categorized as moving animals, people, trees that are swaying, birds, and some other objects. The object tracking method divides the area of interest in a picture, tracks its movements and position, and looks for any blobs. Numerous studies have been conducted recently, but this is due to a variety of difficulties, including shadowing, varying lighting conditions, rapid motion, etc. research in this field is still ongoing. The identification and monitoring of multiple objects in dynamic environments have lately attracted the attention of computer vision researchers. In this study, a tracking algorithm that incorporates the object characteristics mentioned above is reviewed and examined. The purpose of this article is to analyse and evaluate the prior approaches for object recognition and tracking using video sequences over various stages. Deep learning has received a lot of attention lately due to its excellent performance.

Reference :