Abnormal behavior identification method and system based on feature object and human body key point, and medium

The invention discloses an abnormal behavior identification method and system based on a feature object and a human body key point, and a medium. The method comprises the following steps: detecting the position of an object through an FCOS feature object detection method; according to the obtained position of the object, motion prediction is carried out in combination with human body key point information obtained in personnel tool wearing detection; face recognition is carried out in combination with FaceNet; and outputting an alarm result. The method aims at the defects that a traditional human body behavior monitoring platform is high in manpower and material resource cost and slow in response time efficiency, and scene switching and adaptive detection are difficult to rapidly switch when video behavior algorithm research is carried out. The abnormal behavior detection system can accurately detect abnormal behavior events occurring in a monitored area, can timely give out corresponding alarms, can reduce l