In this paper, we develop the Interacting Multiple Model (IMM) estimator using the constant velocity, constant acceleration and coordinated turn model set with Global Nearest Neighbour (GNN) filter to track multiple maneuvering targets. The performance of this IMMGNN filter is further enhanced by incorporating a recursion of the probability of target visibility to take into account instances when a target becomes invisible or when it enters an area occluded by physical barriers or limitations of the sensor. Simulation results demonstrate the superior performance of the IMMGNN filter with target visibility incorporated over its original counterpart.