Reduction of Agents Based on Appearance Probability for Agent-based Human Tracking System

Surveillance systems are widely used in various fields such as businesses and educational institutions. However, as the number of locations and people being monitored increases, it becomes difficult for operators to keep the track of targets. To alleviate the burden on operators, an agent-based human tracking system has been proposed. However, the number of agents in the system increases if tracking accuracy tries to improve. To solve this problem, we propose a method to reduce the number of agents based on appearance probability. By dispatching agents only to locations where the target is likely to appear based on appearance probability, we reduce the number of agents in a system while keeping efficient tracking. Simulation results show that the proposed method can keep a high tracking success rate even with a small number of agents.