Author(s)

V. VIRILI, F. GARZIA & R. CUSANI

Abstract

The purpose of the present work is to find out a new methodology to automatically detect abnormal situations in high risk places through a video surveillance system. The idea is to retrieve true and predicted movement of the people in the scene then, through a classifier, to map out different abnormal situations comparing proper vectors. To reach its purpose, the proposed methodology uses a multidisciplinary approach.

Keywords

Abnormal activities recognizer, automatic video surveillance, pedestrian tracker