The evolving network model provides a better method to deal with the structure of groups, including the formation and real-time update. The box-particle evolution networks probability hypothesis density(PHD) filter for group targets tracking is proposed to improve the increase of computational effort about a sequential Monte Carlo(SMC) PHD filter. The proposed algorithm obtains information of group targets which is combined with BP-PHD and evolving network models, and then feeds back the information to the filter. Consequently, the algorithm realizes the tracking and number estimation of group targets. Comparative experiments show that, the proposed algorithm is more effictive than the SMC-PHD filter, especially in high clutter environments.