Chemical plant worker abnormal behavior identification and early warning method and system

The invention discloses a chemical plant worker abnormal behavior identification and early warning method and system. The method comprises four parts of chemical plant worker abnormal behavior data set establishment, worker skeleton joint point detection, worker abnormal behavior identification and abnormal behavior display and early warning. Establishing an abnormal behavior data set according to abnormal behavior characteristics of chemical plant workers; and carrying out framing on the obtained video data, and processing pictures. Acquiring worker skeleton joint point information by adopting an OpenPose algorithm; the improved STGCN is utilized to train a behavior recognition model, and STGCN hyper-parameters are optimized through an improved AO algorithm to improve model performance; carrying out worker behavior identification by utilizing the trained model; and finally, front-end display and early warning are carried out on the identified abnormal behaviors, so that a supervisor can check and process the