The Gaussian mixture cardinality-balanced multi-target multi-Bernoulli filter (GM-CBMeMBer) always uses standard Kalman or extended Kalman in prediction and updating stages. However, its performance declines greatly when the statistical characteristics of the process noise or measurement noise change abruptly. In order to solve this problem, an improved filtering solution which adopts adaptive fading Kalman technique is proposed. It adaptively adjusts the prediction covariance matrix and then the gain matrix in multi-target filtering process by introducing an adaptive fading factor to restrain the divergence of the filter. Simulation results show that the proposed algorithm evidently reduced the influence of inaccurate modeling for process noise or measurement noise caused by abrupt change of noise characterizes and obtained more stable result in multiple target tracking.