Data deduplication technology has been widely applied in cloud storage systems. Under the premise of ensuring data privacy, how to effectively perform deduplication in semi-trusted cloud storage environments becomes one of the primary issues in cloud computing security. Current schemes rely heavily on online trusted third parties to manage data labels and to keep track of the number of users. The trusted third party plays such a vital role in those schemes that it is indispensable even at the cost of unsatisfying efficiency and potential bottleneck. A verifiable secure data deduplication scheme in cloud storage is proposed, which doesnot require any online trusted third party. The dual-tag scheme based on bilinear mapping is adopted to conduct popularity check. The tag is used to retrieve files without leaking any exploitable information. A modified group signature scheme is designed to prevent the cloud server from forging popularity query results. Users can verify the authenticity of query results from the cloud server. The multi-layered cryptosystem is adopted in the proposed scheme, in which different encryption strategies are applied according to the popularity of specific data. The correctness and security of the proposed scheme are analyzed and proved. Simulation results show that the proposed scheme is secure and efficient.