Research And Implementation Of Fast Identity Registration System Based On Audio-Visual Fusion

In the research of family service robot, face recognition has become a hotspot because of its wide application prospect in home security, human-computer interaction and authentication. The robot must first solve the user's identity registration problem before performing the task of face recognition. To make the registration process more natural and convenient, this paper combines the human audio-visual information, first use the voice interactive way to collect the user's name and face image information, and then use the improved deep neural network and transfer learning method quickly training out the face recognition model to achieve the user's identity registration. The experimental results show that the proposed method not only improves the convenience of identity registration but also shortens the time it takes for user's identity registration.