Topic-Guided Attention for Image Captioning
This work addresses a common defect in image captioning models by improving attention mechanisms for better feature selection, though it is incremental as it builds on existing attention-based methods.
The paper tackles the problem of selecting informative image features in attention-based image captioning by proposing a topic-guided attention mechanism that integrates image topics as higher-level guidance, achieving state-of-the-art performance on the Microsoft COCO dataset.
Attention mechanisms have attracted considerable interest in image captioning because of its powerful performance. Existing attention-based models use feedback information from the caption generator as guidance to determine which of the image features should be attended to. A common defect of these attention generation methods is that they lack a higher-level guiding information from the image itself, which sets a limit on selecting the most informative image features. Therefore, in this paper, we propose a novel attention mechanism, called topic-guided attention, which integrates image topics in the attention model as a guiding information to help select the most important image features. Moreover, we extract image features and image topics with separate networks, which can be fine-tuned jointly in an end-to-end manner during training. The experimental results on the benchmark Microsoft COCO dataset show that our method yields state-of-art performance on various quantitative metrics.