The Internet of Things (IoT) has become the hottest in both the research community and industry. Among them, Radio Frequency Identification (RFID) plays a key role in IoT. On the RFID tags estimation problem, most existing researches are trying to identifying tags' ID rather than counting the number of tags. But the number of tags is useful information in many applications such as stock management and traffic flow management. Massive tags cause taking a lot of cost and time in the estimate. So an essential problem is how to quickly and accurately estimate the number of massive tags. In order to solve this problem, this paper proposes an accuracy and efficiency hybrid scheme by decreasing time and space complexity. The results of simulation conducted to test the effectiveness of the proposed approach, which matches well with the theoretical analytical model.