Cross-modal transformer with language query for referring image segmentation.