Provable data possession (PDP) and proofs of retrievability (POR) are techniques for a client to verify the integrity of outsourced data in cloud storage. Recently, numerous PDP and POR schemes have been proposed while the techniques are widely used in academic and industrial community. However, due to specific and unique requirements of different groups, PDP/POR schemes vary and many functionalities such as data deduplication have not been implemented. How to construct an efficient group PDP/POR scheme to meet these unique requirements of functionality and security has received much attention. In this paper, a group PDP with deduplication (GPDP) is presented. Based on matrix calculation and pseudo-random function, GPDP can efficiently guarantee data possession with deduplication, as well as defend against selective opening attacks of a malicious party.The security of GPDP in the standard model is proved and a prototype based on GPDP scheme in a realistic cloud platform of Baidu is implemented. To evaluate the performance of GPDP, this work utilizes data size of 10GB for experiments and analysis. The result of experiments show that GPDP can guarantee data possession efficiently with deduplication and protect against selective opening attacks. In particular, the performance is superior to private schemes in the phase of pre-process and public schemes in the phase of verification (as efficient as private scheme in the phase of verification). Furthermore, GPDP reduces the extra storage and communication cost to a minimum than the other PDP/POR schemes applied in a group.