The present disclosure provides a method for recognizing a human motion, a method for recognizing a user action and a smart terminal. The method for human motion recognition comprises: collecting human motion data to train to obtain a feature extraction parameter and a template data sequence; in one human motion recognition, collecting data for performing human motion recognition to obtain an original data sequence; using the feature extraction parameter to perform feature extraction on the original data sequence, reducing the number of data dimensions of the original data sequence, and obtaining a test data sequence after the dimension reduction; matching the test data sequence with the template data sequence, and confirming that a human motion corresponding to the template data sequence associated with the test data sequence occurs when a successfully-matched test data sequence exists. By performing dimension reduction on the test data sequence, the present disclosure lowers requirements for human motion postures and cancels noise, then matches the data after the dimension reduction with the template, realizes accurate recognition of human motions while reducing the computing complexity, and improves the user experience.