Several objective models have been proposed for measuring the visibility of video stalls (or video freezes); essential to those models, is an accurate detection of the stall segments. However, most (if not all) proposed methods for video stalls detection ignore the possible existence of static video segments (i.e., segments without any temporal activity), as those corresponding to still images, black frames, or frames without foreground and background motion, that will be erroneously classified as stalls. In this paper, we propose a new algorithm for the automatic detection of static video segments, that allows to discriminate between video stalls, no motion segments, still images and black frames; stall segments are further classified as resulting from dropped frames or delayed frames.