Text this: Visual object segmentation based on temporal and linguistic cues