A Survey on Video Anomaly Detection
Rajesh Kumar Yadav, Rajiv Kumar
Abstract
The number of surveillance cameras has increased considerably over the last decade, and so is the research in order to reduce the human intervention in surveillance in order to automate the process. Because human-based monitoring is time-consuming and difficult, there is an increasing demand for autonomous systems to detect abnormalities in crowded locations. Deep learning has shown to be a game-changing computing technology in the realm of computer vision. As a result, it's routinely utilised to perform complex cognitive tasks like spotting abnormalities in surveillance footage. Deep learning anomaly detection technologies beat conventional machine learning systems. Our research seeks to provide a complete analysis of video anomaly detection systems that employ deep learning that have been published since 2019. In this paper, we examined the learning methods as well as the dataset used for training and testing. Along with it, the model's accuracy is described. In addition, this work gives a brief review of video datasets for anomaly detection. Furthermore, present and emerging trends are emphasised.