Detection of Abnormal Human Behavior Using Unsupervised Learning in Video Surveillance Systems

As public safety demands increase, there is a growing need for intelligent surveillance systems capable of detecting abnormal human behavior in real-world settings. Numerous detection techniques based on machine learning and deep learning models have been developed for abnormal behavior detection. Among these, unsupervised learning enables anomaly detection without labeled data by learning normal patterns and identifying deviations, making it a practical solution to address the shortage of abnormal data. This paper surveys and analyzes recent unsupervised learning techniques for detecting abnormal human behavior in surveillance video streams, reviewing commonly used datasets, discussing practical limitations, and identifying areas for improvement to enhance the reliability and efficiency of unsupervised models in surveillance applications.