An Analysis of Artificial Intelligence Techniques in Surveillance Video Anomaly Detection: A Comprehensive Survey

Abstract

Surveillance cameras have recently been utilized to provide physical security services globally in diverse private and public spaces. The number of cameras has been increasing rapidly due to the need for monitoring and recording abnormal events. This process can be difficult and time-consuming when detecting anomalies using human power to monitor them for special security purposes. Abnormal events deviate from normal patterns and are considered rare. Furthermore, collecting or producing data on these rare events and modeling abnormal data are difficult. Therefore, there is a need to develop an intelligent approach to overcome this challenge. Many research studies have been conducted on detecting abnormal events using machine learning and deep learning techniques. This study focused on abnormal event detection, particularly for video surveillance applications, and included an up-to-date state-of-the-art that extends previous related works. The major objective of this survey was to examine the existing machine learning and deep learning techniques in the literature and the datasets used to detect abnormal events in surveillance videos to show their advantages and disadvantages and summarize the literature studies, highlighting the major challenges.

Introductory Chapter: Anomaly Detection – Recent Advances, AI and ML Perspectives and Applications

“…Patriarca et al [8] delve into the importance of weather forecasting for aerodrome operations and propose a machine learning-based approach for anomaly detection in historical weather data. Finally, Şengönül et al [9] explore the use of AI in surveillance video anomaly detection, noting the increasing need for automated systems due to the sheer volume of video data being generated.…”

Three-Stage Deep Learning Framework for Video Surveillance

“…In this regard, the field of the video-based detection of abnormal situations, as referenced in numerous studies ( [1][2][3][4][5][6][7][8]), is experiencing a surge in interest. Research in this area, especially studies investigating surveillance video anomaly detection (SVAD) using deep learning [9], is progressing rapidly. This research domain has evolved to encompass a broader scope, extending beyond behavior-based detection to include advanced areas like human facial emotion recognition for anomaly detection [10], illustrating the expanding reach and depth of investigations in this field.…”

Multi-Object Tracking Using Kalman Filter and Historical Trajectory Correction for Surveillance Videos

“…Multi-object tracking refers to multi-object tracking is one of the important research topics in the field of computer vision, with the wide application of surveillance equipment and the rapid development of computer vision technology, multi-object tracking technology based on surveillance videos is widely used in the field of traffic control [1], intelligent security [2] and accident early warning [3], which provides support for a large number of practical application scenarios. Through the real-time tracking of pedestrian targets in the surveillance videos, timely detection of accidents, achieving early warning, and reducing the economic losses caused by accidents and pedestrian hazards, multi-object tracking technology has gradually become an important direction for developing security monitoring [4]. However, in practical applications, the target to be tracked is not static, and the main challenges of multi-object tracking include video blurring, the presence of occlusion, and, the sudden disappearance of the target, which makes the tracking algorithm suffer a great impact on the tracking performance [5,6].…”