The invention relates to a video moving object tracking and identity recognition method based on time characteristics, belongs to the technical field of video image processing, and solves the problemthat the existing method is inaccurate in moving object tracking and identity recognition. The method comprises the following steps: constructing a figure detection tracking network, and carrying outtraining; usingthe figure detection tracking network to extract spatial features and time features of figures to obtain moving figure coordinates and identifiers which are in one-to-one correspondencein the time dimension; Inputting continuous sequence frames of the video stream into the trained figure detection tracking network, detecting moving figures in the video, and carrying out real-time tracking according to figure coordinates and matching relationships of the figure coordinates in different frames; And performing face recognition on the detected moving person, and determining the identity of the tracking object. face and character space-time fecatures are extracted by using video continuous framesThe accuracy of face recognition and the accuracy of character detection tracking are greatly improved through the time characteristics, and accurate character tracking and face recognition are achieved on the whole.