The invention discloses a system for realizing voice wake-up free based on mouth key position feature detection. The system comprises a face detection module, a face tracking module, a mouth key pointposition detection module and a mouth key point position feature detection module, wherein the face detection module is connected with the face tracking module, the face tracking module is connectedwith the mouth key point position detection module, and the mouth key point position detection module is connected with the mouth key point position feature detection module. According to the system,the defect of voice man-machine interaction can be overcome. Meanwhile, in combination with the characteristic of man-machine interaction based on special postures, a key point position characteristicmatrix of the mouth is detected through the machine vision technology, and the key point position characteristic matrix is detected through the cyclic convolutional neural network, so that recognition of the action intention of the mouth of the detected person is achieved, and finally voice wake-up-free of the television is achieved.