An Analystical Approach for Survey on Various Techniques for Video Anomaly Detection

Abstract

Detection of irregularities in videos is an issue that has already been researched for over a decade. Because of its broad applicability, this field has attracted the attentions of scholars. As a result, a large variety of approaches have been employed over the years and these strategies variety from matrix factorization to approaches based on machine learning. Several studies are already being carried out in this area, but this paper aims at providing an overview of the recent developments in the field of detection of anomalies using Deep Learning. This study review summarizes research patterns in video feeds of a single scene on the subject of anomaly detection. We address the different formulations of problems, publicly accessible databases, and standards for assessment. In an intuitive taxonomy, we categories and place previous research and provide a detailed comparison of the accuracy of several algorithms on generic test sets. Finally, standard procedures are also given and some potential recommendations for future research are suggested.