ABSTRACT
During this pandemic situation, it is very important to maintain social distance between the people in crowded places. It is impossible for the management of shopping malls, cinema theatres, and other crowded places to observe social distance manually. In this paper, the proposed model studies different recognition algorithms for identifying persons through CCTV surveillance and checks the distance between objects and other persons through object detection algorithms. Object recognition is identifying the content surrounded by the boundary box to make the system get trained on the labels associated with objects. In contrast, object detection is a process of identifying the region in the image where objects exist and marking a rectangular boundary box along with that object. Object recognition is achieved using either machine learning models like a combination of Histogram Gradients and SVM, visual features are grouped as a bag and VJ algorithm, or deep learning models with the help of CNN. There are a few limitations to be taken care of by the recognition. Multi-classification of objects is a complicated process in fully connected neural networks, and the localization process has to identify the coordinates of the boundary box to get more accurate results. The limitations of the object detection process are associated with the computation of geometrical surface areas and volumes. This paper discusses object detection and recognition and their applications and different algorithmic approaches. It also surveyed various researches that are carried out on applications that are developed using these approaches.
