Person-on-Person Violence Detection in Video Data

We address the problem of detecting human violence in image sequences, such as fist fighting, kicking, hitting with objects, etc. To detect violence we rely on motion trajectory information and on orientation information of a person’s limbs. We define an Acceleration Measure Vector (AMV) composed of direction and magnitude of motion and we define jerk to be the temporal derivative of AMV. We present results from several data sequences involving multiple types of violent activities.

 

Video Sequences

Movies Involving Violence
Movies NOT Involving Violence
Intermediate Movies

Related Publication

Ankur Datta, Mubarak Shah, and Niels Lobo Person-on-Person Violence Detection in Video Data, International Conference on Pattern Recognition, Quebec City, Quebec, Aug 11-15, 2002