A human sitting posture sensing terminal is designed based on end-to-cloud combination framework. It is arranged according to the characteristics of pressure distribution of legs and buttocks as human is sitting

the groups of resistive pressure sensor and measurement areas. A composite filtering algorithm is proposed for outliers filtering and smoothing pretreatment. The axial tilting coefficient of front-rear and the left-right are defined to reduce the dimension of pretreated pressure data. A two-layer structure support vector machine（ SVM） algorithm is constructed for multi-classification of sitting posture utilizing the axial tilt coefficients. The multi classifications calculation are achieved and sent down to the sensing terminals by wireless networks. The effectiveness of the proposed algorithm is validated by experimental results.