In this paper, a novel shape and texture error concealment technique for segmented object-based video scenes is proposed. This technique is different from existing concealment techniques because it considers, not only the corrupted video objects to be concealed, but also the context/scene in which they are inserted. In the proposed technique, concealment is done by using information from the current time instant as well as from the past. The obtained results suggest that the use of this technique significantly improves the subjective visual impact of scenes on the end-user, when compared to independent concealment of video objects.