Facial paralysis is a medical entity that a patient losses of ability to move one side of the face. It is necessary to quantitatively assess for early treatment. In Japan, the Yanagihara grading scale is usually used to measure the degree of condition. This paper focuses on the tracking-based approach of interest points to provide evaluation for the Yanagihara grading system. Experiments show that our method outperforms other techniques on motion tasks on dynamic expression database.