Published Paper Details:

Smart CCTV systems rely on abnormal event detection, human behavior detection, and object recognition to enhance surveillance capabilities. These systems can identify abnormal events, detect unusual human behavior, and monitor the state of the environment. Machine vision and machine learning techniques are commonly used to detect and identify anomalies in CCTV video feeds. Typically, these systems process video frames individually using supervised learning to train the algorithms. However, since anomalies can be diverse and difficult to predict, unsupervised and semi-supervised learning methods are being adopted to train the system.By using these techniques, the workload of manual detection and alert creation for abnormal events in CCTV feeds can be minimized or eliminated. Additionally, the storage efficiency of the system is improved by storing abnormal events in their original quality and storing normal scenarios in lower quality for archiving purposes.