Spatio-temporal anomaly detection in crowd movement using SIFT

Crowd behaviour detection is very promising area of research these days in field of computer vision and artificial intelligence. A novel framework based algorithm has been proposed in this research paper which focusses three key elements of moving crowd [i] Dynamics based appearance of the crowd [ii] Temporal - Spatial deviation based anomaly [iii] Intrinsic interaction based deviation flow change. This framework deals with micro as well as macro level anomaly detection using SIFT techniques in structured & Semi-Structured crowd. This proposed framework out performs several existing state of art techniques being used for anomaly detection.