The present invention provides a video conference system that is capable of adjusting image window sizes based on the number of video conference participants included in an image window. In particular, face recognition technology is applied to determine the number of persons included within an image associated with a video call. The information regarding the number of participants in the video call is used as a factor in determining the area of a video output device in which the image from the video call is presented at another video endpoint. The sizing of individual image windows may also be determined based on the total number of participants in a video conference.