Person-on-person Violence Detection in Moving and Stationary Camera Videos.
Ankur Datta, Niels da Vitoria Lobo, Mubarak A. Shah · Journal of International Crisis and Risk Communication Research · 2003
This paper is about detecting human violence in video, such as fist fighting, kicking, etc. Our violence detection algorithm is able to deal with video obtained from both moving and static cameras. We employ motion trajectory information, orientation information of a person's limbs, Acceleration Measure Vector (AMV) information, and jerk information. We present results from several moving and static camera sequences which involve different types of violent actions.