Semantic Movie Scene Segmentation Using Bag-Of-Words Representation