In public areas such as airports and subway stations, where there is a high flow of people, the use of monitoring systems to ensure public safety is of paramount importance. Traditional monitoring methods that rely on manual observation are unable to effectively capture various abnormal behaviors in real-time and with precision. To address this challenge, this project focuses on the detection of abnormal behaviors within intelligent monitoring systems, with a particular emphasis on monitoring passenger behaviors, thereby providing robust support for enhancing the safety management level in public places. The project employs two key technologies: human skeleton recognition and human pose analysis. By accurately locating and tracking skeletal joints in surveillance videos and conducting in-depth analysis of human poses, the system achieves highly accurate recognition of passenger behaviors. It can clearly distinguish between normal and abnormal behavior patterns, providing a reliable basis for subsequent warnings and handling. The system implements several functions, such as passenger loitering warning, which alerts security personnel when a passenger is detected lingering in a specific area for an extended period; left-luggage warning, which triggers an alarm after an item has been unattended for a certain amount of time to prevent loss of property or safety incidents; unauthorized intrusion alarm, which promptly detects and reports unauthorized entry into restricted areas; and fall detection, which accurately identifies when a passenger has fallen and provides timely assistance to those in need. These functionalities significantly enhance the intelligence of the monitoring system and inject strong momentum into the safety assurance efforts in public places.