The increase in the deployment of surveillance camera in outdoor and indoor settings have resulted in a growing demand for intelligent systems that can accurately detect and recognize human actions as well as other entities of interest within the captured video data. Although, human action recognition is a well-established topic in computer vision, abnormal behaviour detection has recently received increased research attention. Several abnormal behaviour detection systems have been proposed over the years to ensure human safety. However, only a few comprehensive and systematic reviews report on the current state and future direction of video anomaly detection(VAD) research. This present effort aims to contribute a systematic and detailed review of current research and advances in the detection of anomalous actions and entities in videos. The review focuses on studies published between 2003 and 2023. During the literature selection process, 530 scholarly articles were identified and evaluated to showcase prevalent research trends, techniques, datasets, and frameworks within the realm of VAD This review aims to offer a comprehensive understanding of key areas of focus among researchers, provide resources for commonly used public datasets for evaluation and experimentation, and examine advancements and integration of network design to accommodate the needs for handling multimedia information. To sum up, this study highlights various potential opportunities and obstacles about the VAD domain.VAD has gained significant interest in recent years due to its potential applications in various domains such as security, surveillance, traffic monitoring, medical imaging among others. Areas of further research include, Anomaly detection in real-time, multi-camera anomaly detection, privacy preserving in VAD to mention a few. The study revealed some key trends and challenges in VAD that can guide future research direction.

中文翻译：

---

视频异常检测：问题和前景的系统回顾


室外和室内环境中监控摄像头部署的增加导致对能够准确检测和识别人类行为以及捕获的视频数据中其他感兴趣实体的智能系统的需求不断增长。尽管人类动作识别是计算机视觉领域的一个成熟主题，但异常行为检测最近受到了越来越多的研究关注。多年来，已经提出了几种异常行为检测系统来确保人类安全。然而，只有少数全面、系统的综述报告了视频异常检测（VAD）研究的现状和未来方向。目前的工作旨在对视频中异常行为和实体检测的当前研究和进展进行系统和详细的回顾。该综述重点关注 2003 年至 2023 年间发表的研究。在文献筛选过程中，确定并评估了 530 篇学术文章，以展示 VAD 领域内的流行研究趋势、技术、数据集和框架。该综述旨在提供对研究人员关注的关键领域，为常用的公共数据集提供资源以进行评估和实验，并检查网络设计的进步和​​集成，以满足处理多媒体信息的需求。综上所述，本研究强调了 VAD 领域的各种潜在机遇和障碍。近年来，VAD 由于其在安全、监控、交通监控、医学成像等各个领域的潜在应用而引起了人们的极大兴趣。 进一步研究的领域包括实时异常检测、多摄像头异常检测、VAD 中的隐私保护等。该研究揭示了 VAD 的一些关键趋势和挑战，可以指导未来的研究方向。