Bridging by Word: Image Grounded Vocabulary Construction for Visual Captioning

37011

Knowledge Triples