Target Trajectory Prediction in PTZ Camera Networks

In this paper we compare different mechanisms for the prediction of targets position inside a PTZ camera network. The goal is to predict the next location of each target with higher accuracy, in order to better plan the movements of the cameras at the next time step. For that purpose, we are proposing a probabilistic multimodal approach, and show that this prediction method can improve the total coverage of a camera network compared to other probabilistic prediction methods.