Detecting insecure person in crowd with human sensor

We consider the problem of detecting potential dangerous people (i.e. terrorist) in a crowded but closed environment such as airport (and railway station). Instead of conventional sensor, we proposed to use human (passengers) themselves as sensors to detect and identify the target, processing the information from environment with human brain and their cognitive capability. Passengers report the identified issues as tweets with their mobile phone. The monitoring system collects the tweets and classifies their creditability through semantic analysis and learning. Then a temporal model is applied to compute the overall credibility of event. After the event is identified, photos and text description in tweets are combined with sensors (like camera) to identify and track the target people with range image motion detection algorithm. The work is currently in progress of designing model and prototype system