Abstract: The number of on-road traffic accidents is escalating every day and a system to monitor these and take required actions is the need of the hour. This paper proposes a model that can effectively detect accidents and report the part of the traffic video containing the accident to the nearest set of hospitals. It uses Gaussian Mixture model to detect vehicles and mean shift algorithm to track their movement, upon which an accident is presumed to occur based on the values of three parameters of the vehicle; position, acceleration and direction of motion of the vehicle. If the thresholds of the aforementioned values are exceeded and accident is detected, then the video in the time frame of the accidents is rendered to the nearest registered hospital.

Keywords: Gaussian Mixture Model, Mean shift algorithm, Accident Detection Algorithm

| DOI: 10.17148/IJIREEICE.2019.7303