Text this: Keyword-Conditioned Image Segmentation via the Cross-Attentive Alignment of Language and Vision Sensor Data.