A Real-time, Multi-space Incident Detection System For Indoor Environments
Free (open access)
Volume 8 (2018), Issue 2
266 - 275
DIMITRA TRIANTAFYLLOU, STELIOS KRINIDIS, DIMOSTHENHS IOANNIDIS & DIMITRIOS TZOVARAS
A real-time, multi-camera multi-space incident detection system for indoor environments is proposed. The system focuses on industrial environments and recognizes incidents such as human falls, intrusions to restricted areas and collisions. Building Information Models (BIM) provide all the necessary infor- mation for the set up of the system. Furthermore, once an incident is detected an alarm notifying of its type and exact location on the building is triggered so that the appropriate coping mechanism can hasten promptly to the point of the incident. Based on building information about each area particularities and special conditions, e.g. moving machinery or chemical agents, the appropriate coping measurements can be taken. Moreover, the use of the incident detection system for a long period can provide statistics about the inherent dangers of the building. Experimental results seem very promising.
collision, fall, incident detection, industrial environment, intrusion.