Disclosed are various embodiments to automatically generate network pages from extracted media content. In one embodiment, it is determined that a first facial expression of a face appearing in a frame of a digitally encoded video matches a second facial expression specified in a media extraction rule. The frame of the digitally encoded video is selected in response to the determination that the first facial expression matches the second facial expression. A user interface is generated that includes an image extracted from the selected frame of the digitally encoded video.