In a processing method for a video conference, start speaking events and end speaking events of endpoints joined in the video conference are detected. Video streaming of a speaking endpoint is displayed in a speaking area of a foreground window of a display device. A current time “t”, a duration time “T” of the video conference, a start speaking time “si” and an end speaking time “ei” of the endpoints are recorded. Real-time activity scores of each of the endpoints are calculated and updated according to video recording contents. Video streaming of non-speaking endpoints are displayed in corresponding areas of the display device according to the calculated activity scores.