Fig. 3From: Image captioning model using attention and object features to mimic human image understandingComparison between our baseline model (without object features) and our proposed model. a Results on MS COCO testing set. b Results on MS COCO development setBack to article page