Auto-Detection of Unusual Events in Metro Stations via Security Cameras

Dagli M. A. , Erdem Ç.

21st Signal Processing and Communications Applications Conference (SIU), CYPRUS, 24 - 26 Nisan 2013 identifier identifier


Due to the excess of the security cameras on the metro stations; staff is unable to track all of the cameras at same time. So that he may miss many of the important events. In order to detect these events automatically and generate warnings to the staff for preventing events before dangerous results occurred is the main goal of the research. With this study functions like train detection, yellow line intrusion, fallen passenger detection, restricted area violation and open door detection are realized. According to the 200 hours tests system reaches more than 99% success rate about detecting events.