Text-Guided Attention Model for Image Captioning

Jonghwan Mun, Minsu Cho, Bohyung Han

Proceedings of the AAAI Conference on Artificial Intelligence · 2017 · 90 citations · 26 references

DOIFull text

Open access

Concepts

TL;DR

Visual attention is crucial for image understanding and has proven effective in generating natural language descriptions, while recent studies indicate that associated language can steer visual attention during cognition. The study introduces a text‑guided attention model for image captioning that learns to drive visual attention using associated captions. The model employs an exemplar‑based learning approach that retrieves associated captions from training data for each image and uses them to learn attention over visual features. The model achieves state‑of‑the‑art performance on the MS‑COCO Captioning benchmark and improves scene description by accurately distinguishing small or confusable objects.

Abstract

Visual attention plays an important role to understand images and demonstrates its effectiveness in generating natural language descriptions of images. On the other hand, recent studies show that language associated with an image can steer visual attention in the scene during our cognitive process. Inspired by this, we introduce a text-guided attention model for image captioning, which learns to drive visual attention using associated captions. For this model, we propose an exemplar-based learning approach that retrieves from training data associated captions with each image, and use them to learn attention on visual features. Our attention model enables to describe a detailed state of scenes by distinguishing small or confusable objects effectively. We validate our model on MS-COCO Captioning benchmark and achieve the state-of-the-art performance in standard metrics.

References

26