This paper proposes an intelligent hotel management system based on facial recognition. Its facial recognition function is completed based on the OpenCV open source computer vision library. Through the research of AdaBoost algorithm, Camshift algorithm and deep confidence network combined with fuzzy theory, this paper proposes a new hotel occupancy management system, which realizes the integration of person and witness and saves energy and electricity. At the same time, in view of the shortcomings of new smart hotels in resource conservation, this paper proposes a new human body detection method, which achieves the effect of supplying power when people are present and cutting off power when people leave. Through this research, the dependence of traditional human sensors on scenes is solved, and the detection reliability is improved.