Abnormality analysis for densely-populated public places based on Bayesian inference forecasting

Abstract

Gathering is an important sign of social security event in public places. Timely detection of abnormal gathering contributes to effective warning, thereby avoiding the occurrence of serious events. In this paper, we identify abnormal situation according to a reasonable safe range based on people flow characteristics of public places. We employ Bayesian inference forecasting approach to cope with emergency forecasting and early warning.