A method to automatically classify video content retrieved over the Internet and, in particular, to determine whether a candidate video is an instance of a source video. The method begins by sampling static image frames from each of the candidate video and the source video. Based on the sampled static image frames, a set of static image frames in the candidate video that match a corresponding set of static image frames in the source video are then identified. Metadata associated with the sets of matching static image frames are then compared to determine how closely a temporal order of the set of static image frames in the candidate video matches the temporal order of the set of static image frames in the source video. Based on a measure of temporal order match, a classification is then applied to the candidate video to complete the method.
展开▼