Recognizing anomalous conduct in crowded environments quickly and automatically can greatly improve public safety.Real-time surveillance systems are in high demand as urbanization and industrialization spread rapidly. Because of their reliance on artificial intelligence, anomaly identification systems only tackle some of the challenges, mainly overlooking the changing nature of abnormal behavior over time. Anomaly identification techniques also have the additional issue of requiring a training dataset with established normalcy and known error levels. Common methods for spotting anomalies on the WoT platform include kee ping tabs on user behavior and using visual frames to describe crowd features like density, direction, and motion pattern. Real -time security monitoring based on the WoT platform and machine learning algorithms would, thus, greatly improve the influential detection of abnormal crowd actions.