Paying Attention to Descriptions Generated by Image Captioning Models
This work addresses the gap in understanding how humans and machines describe images, with incremental insights into attention mechanisms in captioning models.
The study investigated the relationship between visual saliency and object mentions in image descriptions, finding that humans mention salient objects earlier and that better-performing captioning models align more with human attention patterns. The proposed saliency-boosted model did not significantly improve performance on the MS COCO dataset but showed better generalization on unseen data.
To bridge the gap between humans and machines in image understanding and describing, we need further insight into how people describe a perceived scene. In this paper, we study the agreement between bottom-up saliency-based visual attention and object referrals in scene description constructs. We investigate the properties of human-written descriptions and machine-generated ones. We then propose a saliency-boosted image captioning model in order to investigate benefits from low-level cues in language models. We learn that (1) humans mention more salient objects earlier than less salient ones in their descriptions, (2) the better a captioning model performs, the better attention agreement it has with human descriptions, (3) the proposed saliency-boosted model, compared to its baseline form, does not improve significantly on the MS COCO database, indicating explicit bottom-up boosting does not help when the task is well learnt and tuned on a data, (4) a better generalization is, however, observed for the saliency-boosted model on unseen data.