The embodiment of the invention discloses a face recognition method and device based on a video sequence. One specific execution mode of the method comprises the steps that responding to detection ofa face object contained in image frames in the acquired video sequence, the face object is tracked to extract multiple target image frames containing the face object from the video sequence; featuresof the face object contained in the target image frames are extracted; and matching with a preset face feature library is performed based on the extracted features, and identification information of the face object is obtained, wherein the preset face feature library comprises face features of multiple preset user objects marked with identification information. Through the execution mode, recognition of the face object in videos is realized, and the accuracy of face recognition and the efficiency of video content property determination are improved.