A motion recognition-based interaction method includes the steps of: providing a first motion image as a reference to a user terminal, obtaining a second motion image of the motion of a user from the user terminal, extracting a reference motion from the first motion image, extracting the user motion from the second motion image, and evaluating the motion state of the user by comparing the reference motion with the user motion, and providing feedback to the user based on the evaluation of the motion state. The feedback includes visual feedback.